Methodology for Assessing Risks to Ship Traffic from

Transcription

Methodology for Assessing
Risks to Ship Traffic from
Offshore Wind Farms
Photo: Hans Blomberg +46 70 550 0121
VINDPILOT-Report
to
Vattenfall AB
and
Swedish Energy Agency
SSPA Sweden AB
SSPA Sweden AB
12 June 2008
POSTADRESS
POSTAL ADDRESS
BESÖKSADRESS
STREET ADDRESS
TELEFON
TELEPHONE
TELEFAX
TELEFAX
E-MAIL
E-MAIL
ORG NR
REG NO.
BANKKONTO
BANK ACCOUNT
BANKGIRO
BANK GIRO
BOX 24001
SE-400 22 GÖTEBORG
SWEDEN
CHALMERS TVÄRGATA 10
GÖTEBORG
SWEDEN
NAT 031 - 7729000
INT +46 - 31 7729000
NAT 031 - 7729124
INT +46 - 31 7729124
[email protected]
556224-1918
SE-BANKEN
5027-1002190
152-4875
SSPA Sweden AB
POSTADRESS
POSTAL ADDRESS
BESÖKSADRESS
STREET ADDRESS
TELEFON
TELEPHONE
TELEFAX
TELEFAX
E-MAIL
E-MAIL
ORG NR
REG NO.
BANKKONTO
BANK ACCOUNT
BANKGIRO
BANK GIRO
BOX 24001
SE-400 22 GÖTEBORG
SWEDEN
GÖTEBORG
SWEDEN
NAT 031 - 7729000
INT +46 - 31 7729000
NAT 031 - 7729124
INT +46 - 31 7729124
[email protected]
556224-1918
SE-BANKEN
5027-1002190
152-4875
REPORT
Subject
Report
Methodology for Assessing Risks to Ship Traffic from
Offshore Wind Farms
2005 4028
Project manager
Björn Forsman
Customer/Contact
Authors
Göran Loman
Director, Project Development, Vattenfall AB
Joanne Ellis
Björn Forsman
Johannes Hüffmeier
Jessica Johansson
Order
Date
2006-03-03
2008-06-12
This report summarise the findings of a research project funded by Vattenfall and the
Swedish Energy Agency and conducted by SSPA Sweden AB. The project addresses
methodologies for risk assessment concerning navigational risks to ship traffic from offshore
wind farms and aims to develop a consistent and conclusive approach for risk based decision
making.
Jim Sandkvist
Vice President
Björn Forsman
Project Manager
SSPA Sweden AB
POSTADRESS
POSTAL ADDRESS
BESÖKSADRESS
STREET ADDRESS
TELEFON
TELEPHONE
TELEFAX
TELEFAX
E-MAIL
E-MAIL
ORG NR
REG NO.
BANKKONTO
BANK ACCOUNT
BANKGIRO
BANK GIRO
BOX 24001
SE-400 22 GÖTEBORG
SWEDEN
GÖTEBORG
SWEDEN
NAT 031 - 7729000
INT +46 - 31 7729000
NAT 031 - 7729124
INT +46 - 31 7729124
[email protected]
556224-1918
SE-BANKEN
5027-1002190
152-4875
TABLE OF CONTENTS
SUMMARY ........................................................................................................................6
ACKNOWLEDGEMENTS ..............................................................................................9
1
INTRODUCTION ............................................................................................11
1.1
1.2
1.3
1.4
BACKGROUND ....................................................................................................11
SCOPE .................................................................................................................12
METHODOLOGY ..................................................................................................13
RISK ASSESSMENT ..............................................................................................13
2
REVIEW OF CURRENT METHODS/GUIDELINES AND STATE OF
THE ART ..........................................................................................................17
2.1
2.1.1
2.1.2
2.1.3
2.1.4
2.1.5
2.1.6
2.1.7
2.1.8
2.1.9
2.1.10
2.1.11
2.1.12
2.2
2.2.1
2.2.2
2.2.3
NATIONAL GUIDELINES ......................................................................................17
Belgium ..............................................................................................................17
Canada ................................................................................................................17
Denmark .............................................................................................................18
France .................................................................................................................18
Germany .............................................................................................................19
Ireland.................................................................................................................21
The Netherlands .................................................................................................22
Norway ...............................................................................................................22
Spain ...................................................................................................................23
Sweden ...............................................................................................................23
United Kingdom .................................................................................................26
USA ....................................................................................................................28
INTERNATIONAL RESEARCH PROGRAMMES AND HARMONISATION OF
NATIONAL ASSESSMENT SCHEMES ....................................................................29
Safeship ..............................................................................................................29
Safety at Sea project ...........................................................................................30
Harmonisation discussions in Germany .............................................................31
3
NAVIGATIONAL RISK ASSESSMENT METHODOLOGY ....................32
3.1
3.2
3.3
3.4
3.4.1
3.4.2
3.4.3
3.4.4
3.4.5
3.4.6
3.5
3.6
3.7
INITIAL QUALITATIVE ASSESSMENT ...................................................................32
DEFINITION OF TYPES OF RISKS TO BE CONSIDERED .........................................32
DATA SOURCES AND INPUTS TO ANALYSES.......................................................34
ANALYSIS METHODS ..........................................................................................37
Ship – Wind Farm Collision Probability Estimation ..........................................37
Ship – Wind Turbine Structure Consequence Estimation Techniques...............55
Ship to Ship Collision Probability Estimate .......................................................62
Ship to Ship Collision Consequence Estimate ...................................................63
Grounding Probability Estimate .........................................................................63
Grounding Consequence Estimate .....................................................................64
EFFECTS ON RADAR, RADIO, NAVIGATION EQUIPMENT, ETC. ...........................64
EFFECTS ON SEARCH AND RESCUE OPERATIONS ...............................................65
VALIDATION OF RISK ASSESSMENT/QUALITY
CONTROL/UNCERTAINTY/SENSITIVITY ..............................................................66
Model accuracy and uncertainty.........................................................................67
RISK ACCEPTABILITY AND RISK ACCEPTANCE CRITERIA .................................67
Individual and societal risk.................................................................................68
3.7.1
3.8
3.8.1
4
SSPA REPORT NO:
2005 4028
AUTHORS:
Joanne Ellis, Björn Forsman, Johannes Hüffmeier, Jessica Johansson
3.8.2
3.8.3
3.8.4
3.8.5
3.9
Individual risk acceptance in shipping industry .................................................69
Different types of criteria for offshore wind farms ............................................69
Acceptance criteria in Germany .........................................................................69
Other acceptance criteria ....................................................................................72
STATEMENTS AND RECOMMENDATIONS FROM OTHER STAKEHOLDERS ............74
4
RISK REDUCTION MEASURES ..................................................................75
4.1
4.2
BONN AGREEMENT COUNTER-POLLUTION RECOMMENDATIONS......................79
EVALUATION OF RISK REDUCTION MEASURES..................................................80
5
CASE STUDY: APPLICATION OF SELECTED METHODS TO
KRIEGERS FLAK PROPOSED PROJECT ................................................81
5.1
5.2
5.2.1
5.2.2
5.2.3
5.2.4
5.2.5
5.2.6
5.2.7
5.3
5.4
5.4.1
5.4.2
5.4.3
5.4.4
5.4.5
MODELS ..............................................................................................................82
INPUT DATA ........................................................................................................85
Description of the wind farm and the ship traffic...............................................85
Climate ...............................................................................................................88
Self repair and Emergency anchoring and salvage.............................................90
Position on shipping lane ...................................................................................92
Course offset.......................................................................................................92
Course deviation .................................................................................................93
Causation factor and onboard crew reaction ......................................................93
RESULTS .............................................................................................................95
CONCLUSIONS AND DISCUSSION ........................................................................95
Drifting collision ................................................................................................95
Powered collision ...............................................................................................96
Historical empirical accident statistics ...............................................................97
Consequences .....................................................................................................98
Influence of Kriegers Flak II on the calculations ...............................................98
6
RECOMMENDATIONS AND DISCUSSION ..............................................99
6.1
6.2
6.3
RISK ASSESSMENT METHODOLOGY ....................................................................99
CALCULATION MODELS ...................................................................................100
FOLLOW UP ACTIVITIES ....................................................................................101
7
REFERENCES ...............................................................................................102
APPENDICES
Appendix A
Appendix B
Appendix C
Appendix D
Summary Information on Existing Offshore Wind Farms
Accident Statistics from the Swedish Maritime Safety Inspectorate
Case Study Wind Farm Kriegers Flak: Some detailed information
SSPA Calculation Model for Collision Ship – Offshore Wind Farm
5
SSPA REPORT NO:
2005 4028
AUTHORS:
SUMMARY
As the number of offshore wind farms continues to grow, it becomes apparent that
there is a need for comprehensive assessment guidelines and recommendations to
ensure safe maritime operations and to protect the marine environment. Risk
analyses for estimation of the risks associated with wind farm establishment and
for the identification of relevant risk reduction measures have been conducted for
many projects but the results are sometimes difficult to assess and compare
because different methodologies are used and because there is a lack of
established evaluation criteria. In some countries, governmental agencies and
other organisations have tried to establish harmonised risk assessment methods
and formulate guidelines for the performance of risk analysis of offshore wind
farms and their potential impact on maritime safety. Geographical, environmental
and navigational conditions as well as the permit process differs in different
countries and regions and the wind power industry as well as the competent
authorities in Sweden have identified a need to investigate the current
international state-of-the-art and to develop and establish relevant guidelines to be
applied for offshore wind park projects in the waters around Sweden.
SSPA Sweden AB has extensive experience of maritime and navigational risk
assessment including a large number of navigational risk assessments for offshore
wind farms for various Swedish wind energy companies. In 2005, Vattenfall AB
applied for and was granted financial support of 40% from the Swedish Energy
Agency and commissioned SSPA to conduct the present study. The remaining
60% is funded by Vattenfall. In addition, some of the material presented in this
report has been developed in cooperation with SSPA internal research projects.
The objectives of this study were to provide recommendations for a methodology
for assessing risks resulting from ship navigation in the vicinity of offshore wind
farms, where Kriegers flak would serve as a reference for comparing and
evaluating selected techniques. Methodology used includes collection and review
of literature and published material on wind farm risk assessment, comparative
studies on collision risk models, and regulatory guidelines; information collection
and discussion through email and telephone contact with international experts on
offshore wind farms; consultation and discussion with the project reference group;
case study analysis to assess and compare collision probability methods for
assessing risks of wind farm / ship collisions.
Conclusions and recommendations in general regarding risk assessment
methodology and in detail regarding calculation models are presented and
discussed below.
6
SSPA REPORT NO:
2005 4028
AUTHORS:
It is important that the calculation models are transparent. The intention with the
model developed by SSPA (see Appendix) is that all information about the model
should be explicitly stated. This includes the model structure as well as the input
data. The structure of the SSPA calculation model is similar to other models used
for wind farms and offshore platforms. However, there are models using
simulations (e.g. Monte Carlo simulations) but in the SSPA model no simulations
are used since these make the model less transparent. It is questionable whether
simulations give more accurate results of a risk analysis. The SSPA model is
designed to be simple and transparent, which gives a good prerequisite for
explaining the physics behind the model.
A German harmonisation process has laid a basis for a common harmonised set of
parameters which should be used in risk calculations. However, one should be
attentive to that the process has a set of models as a basis and there may be
recommendations that are valid only for these models and can therefore not be
used commonly. Harmonisation processes such as the German one also requires
transparency in order to give recommendations about for example input data.
Harmonisation can be a natural step to take when the models are presented in
detail. However, as shown in chapter 2, the conditions in the different EUmember states vary a lot and each country may identify and prioritise various
safety aspects differently, and total harmonisation may be difficult. The pilot site
for this project, Kriegers Flak, may serve as an illustration of the need for
harmonisation and bilateral/international assessment discussions.
If several wind farms are planned in the area, cumulative effects on the risk should
be studied. This may require cooperation between different countries. One
example is the proposed Swedish and German parks at Kriegers Flak that are
close neighbours, but are processed separately without consideration of
cumulative effects, while other more distant wind farms on the German side are
considered from an interaction point of view with Kriegers Flak.
Collision frequency models are in general sensitive to changes of certain
assumptions. They also contain an amount of uncertainties. Calculated results in
absolute terms should therefore be carefully interpreted. One way of doing this is
to make relative comparisons instead of using absolute values of acceptance
criteria. If acceptance criteria should be used, it should be stated for which type of
calculation model and with which input data these criteria are valid. One
important relative comparison is a zero-alternative discussion where the
navigational risk in a specific area is compared quantitatively with and without the
presence of the wind park. Comparative studies of the calculated collision
frequency of different traffic lanes can also be applied in order to identify which
ones that stands for the largest contribution.
The aim for the SSPA model is to be as clear as possible concerning sensitivity/
uncertainty. This openness makes the SSPA model more useful and shows the
way to improvements of the model. It has for example become obvious during the
progress of this research project that the function describing the probability that
the crew onboard is not able to react in time to correct the navigational error
7
SSPA REPORT NO:
2005 4028
AUTHORS:
(onboard crew reaction) needs to be further investigated together with the
causation factor. One way of doing this would be possible if the processing of
recorded AIS-data could be further developed.
Another way of relating the results of a risk assessment/analysis is to put it in an
economic context. Cost-benefit analysis is not included in this research project but
could be an interesting task for future projects.
Example of risk reduction measures are presented in this report. Measures that are
associated with low economic costs should always be considered even if the
estimated risk is low. If the estimated risk is high, also more expensive measures
must be considered.
Accident preparedness includes various safety measures but should also be linked
to a control program. One of the objectives with establishing and follow a control
program is that the risk and safety issues will be continuously checked and
updated during the whole life time of the wind farm.
8
SSPA REPORT NO:
2005 4028
AUTHORS:
ACKNOWLEDGEMENTS
ACKNOWLEDGEMENTS
This project has been conducted within the “Vindpilot” framework for support of
technical development and market introduction of wind power energy production
in Sweden. The Vindpilot programme is based on a government decision and the
financial support is administered by the Swedish Energy Agency. Large scale
offshore wind farms and wind energy projects located in the northern fjeld regions
are specifically addressed by the support programme. Since the start of the
programme in 2003 the Energy Agency has supported six large projects and this
project, focussing on risk assessment methodology for offshore wind parks with
the Kriegers flak park as a demonstration site, is one of the six projects supported.
Sweden Offshore Wind AB, a daughter company to Vattenfall AB, is planning to
build 128 wind turbines on Kriegers Flak, about 30 km south of Trelleborg.
Vattenfall outlined the project on risk assessment methodology in cooperation
with SSPA Sweden AB and applied for support from the Vindpilot programme.
According to the programme’s conditions for support, a 40% grant is supplied
from the Energy Agency and the remaining 60% is to be funded by the applicant
(Vattenfall). In addition, some of the material presented in this report has been
developed in cooperation with SSPA internal research projects (see Johansson
(2007) and Johansson et al (Ongoing project)).
The project work started in 2006, when Vattenfall contracted SSPA to prepare and
conduct the project work. A project reference group was established and about
five meetings were arranged. The reference group provided key input and advice
throughout the project. This reference group consisted of the following:
Sven-Åke Blomén, Master Mariner/Senior Administrative Officer at the Swedish
Maritime Safety Inspectorate
Fredrik Dahlström, Project Manager at the Swedish Energy Agency
Amelie Gustafsson Fürst, M.Sc. Eng., Consultant Windpower, Vattenfall Power
Consultant AB
Mats Hörström, Nautical Administrator at the Swedish Maritime Safety
Inspectorate
Anders Lancing, Pilot/Master Mariner/OSC at the Swedish Maritime
Administration
Göran Loman, Ph.D., Director, Project Development at Vattenfall
Markus Lundkvist, Ph.D., Risk Analyst at the Swedish Maritime Administration
Thomas Stalin, B.Sc., Director, Project Development at Vattenfall
9
SSPA REPORT NO:
2005 4028
AUTHORS:
ACKNOWLEDGEMENTS
The participation of the group, through attendance at meetings, and provision of
reference material, comments, and valuable insight, is greatly appreciated. The
picture below shows some of the members in the project team and in the reference
group.
Figure A.1. Picture from reference group meeting June 2nd, 2006.
Back row from the left: Fredrik Dahlström (Swedish Energy Agency), Thomas Stalin (Vattenfall),
Jim Sandkvist (SSPA), Markus Lundkvist (Swedish Maritime Administration).
Front row from the left: Joanne Ellis (SSPA), Sven-Åke Blomén (Swedish Maritime Safety Inspectorate),
Jessica Johansson(SSPA), Amelie Gustafsson Fürst (Vattenfall), Björn Forsman (SSPA).
10
SSPA REPORT NO:
2005 4028
AUTHORS:
Introduction
1
INTRODUCTION
1.1
Background
Wind energy is currently one of the fastest growing sources of energy worldwide,
with annual production increases of about 29% per year in recent years (Earth
Policy Institute, 2006). The Vindeby wind farm in Denmark, built in 1991, was
the first offshore wind energy facility to be built in the world. As of 2007, seven
countries (six in Europe) had operational offshore wind farms (see Appendix for a
summary of offshore wind farms). There are also many offshore wind farms under
construction and in the planning and permitting process. The size of turbines
installed continues to increase – the Beatrice Wind Farm Demonstrator Project in
Moray Firth will install two 5 MW turbines in Moray Firth, off the coast of
Scotland. These turbines will be installed in 45 m deep water (Talisman Energy,
2004).
As the number of offshore wind farms continues to grow, it becomes apparent that
there is a need for comprehensive assessment guidelines and recommendations to
ensure safe maritime operations and to protect the marine environment. One of the
phases in the permit process for new offshore wind farms is consultation and
consideration of the project with the maritime safety authorities. The most
important issue for these authorities is the location of the farm, its impact on ship
traffic and the potential hazards of ship collisions with the wind turbine structures.
Risk analyses for estimation of the risks associated with wind farm establishment
and for the identification of relevant risk reduction measures have been conducted
for many projects but the results are sometimes difficult to assess and compare
because different methodologies are used and because there is a lack of
established evaluation criteria.
In some countries, governmental agencies and other organisations have tried to
establish harmonised risk assessment methods and formulate guidelines for the
performance of risk analysis of offshore wind farms and their potential impact on
maritime safety. Geographical, environmental and navigational conditions as well
as the permit process differs in different countries and regions and the wind power
industry as well as the competent authorities in Sweden have identified a need to
investigate the current international state-of-the-art and to develop and establish
relevant guidelines to be applied for offshore wind park projects in the waters
around Sweden.
11
SSPA REPORT NO:
2005 4028
AUTHORS:
Introduction
SSPA Sweden AB has extensive experience of maritime and navigational risk
assessment including a large number of navigational risk assessments for offshore
wind farms for various Swedish wind energy companies. In 2005, Vattenfall AB
applied for and was granted financial support of 40% from the Swedish Energy
Agency and commissioned SSPA to conduct the present study. The remaining
60% is funded by Vattenfall. In addition, some of the material presented in this
report has been developed in cooperation with SSPA internal research projects
(see Johansson (2007) and Johansson et al (Ongoing project)).
1.2
Scope
farms, where Kriegers flak would serve as a reference for comparing and
evaluating selected techniques.
The risk components covered in this methodology are related to ship operation in
the vicinity of an offshore wind park, and include:
•
Ship navigation and probability of an accident or incident:
current situation, and change in probabilities resulting from the
offshore wind farm
•
Consequences resulting from ship-related accidents and incidents:
environmental consequences, consequences to the ship and personnel,
third-party consequences
All ship types, including commercial vessels, fishing boats, and pleasure craft will
be covered to some extent in this report, although the focus will be on commercial
vessels.
Other issues to be addressed in the study include:
•
Effects on Search and Rescue operations and effects on oil spill
monitoring, surveillance, and response.
•
Risk reduction measures.
12
SSPA REPORT NO:
2005 4028
AUTHORS:
Introduction
1.3
Methodology
The research study included the following main components:
•
collection and review of literature and published material on wind farm
risk assessment, comparative studies on collision risk models, and
regulatory guidelines
•
information collection and discussion through email and telephone contact
with international experts on offshore wind farms
•
consultation and discussion with the project reference group
•
case study analysis to assess and compare collision probability methods
for assessing risks of wind farm / ship collisions
Literature reviewed for the study included navigational risk assessments
completed for existing wind parks, published papers, regulatory documents and
guidelines, and reports from international research projects.
A case study approach was used as part of the investigation and development of
recommendations for collision probability analysis. Data for the Kriegers Flak
wind park site was used to undertake a comparison of calculation models used by
MARIN and GL. The method used for the case study was to simulate or emulate
the two models with SSPA’s model as a starting point.
1.4
Risk Assessment
Risk assessment is a process for identifying and analysing undesirable events or
results of a process, and determining whether the risks are acceptable. If risks are
unacceptable, the process may include recommendations and assessment of risk
control measures. The process can include the following steps:
•
Description of activity or process
•
Hazard identification
•
Accident and Incident Scenario generation
•
Frequency estimation
•
Consequence Estimation
•
Risk evaluation
13
SSPA REPORT NO:
2005 4028
AUTHORS:
Introduction
Further steps can include generation of risk control measures, and a repeat of the
steps to evaluate the potential risk reduction resulting from implementation of the
risk control measures.
In order to make the output of the risk assessment useful for decision making
there is also a need for risk acceptance criteria to guide decision makers to be
consistent in the permit processes.
Risk assessment is used in many industries, and although the steps are similar,
there can be variations to reflect specific industry concerns and focus. The
International Maritime Organization (IMO) has developed a specific risk
assessment process, which is referred to as a Formal Safety Assessment, to be
used in the IMO rule-making process. Guidelines for this process were approved
by the Maritime Safety Committee and the Marine Environmental Protection
Committee in 2001 and 2002 (IMO, 2002). The IMO describes the FSA as “a
rational and systematic process for assessing the risks related to maritime safety
and the protection of the marine environment and for evaluation the costs and
benefits of IMO’s options for reducing these risks”. The steps of FSA are briefly
summarised in the following figure.
14
SSPA REPORT NO:
2005 4028
AUTHORS:
Introduction
What is FSA?
Formal Safety Assessment, FSA is a proactive process introduced by the IMO
(International Maritime Organization) to be used as a tool in the rulemaking process –
it is “one way of ensuring that action is taken before a disaster occurs”. The FSA
preferably addresses a specific category of ships or navigational area but may also
be applied to a specific maritime safety issue to identify cost effective risk reduction
options. The FSA process includes five basic steps:
1. Hazard Identification
List of accident scenarios
2. Risk Analysis
Probability and Consequences
3. Risk Control Options
4. Cost Benefit Assessment
5. Recommendations
More information on the FSA process:
Guidelines for Formal Safety Assessment (FSA) for use in the IMO rule-making
process (MSC/Circ.1023 -MEPC/Circ.392)
Figure 1.1. Formal Safety Assessment Procedure
15
SSPA REPORT NO:
2005 4028
AUTHORS:
Introduction
There are a number of tools for carrying out the “risk analysis” step of a risk
assessment or safety assessment procedure. Tools that have been commonly used
as part of marine risk assessments include:
•
Fault tree analysis
•
Event tree analysis
•
Bayesian network analysis
Fault tree and event tree analysis are two of the main methods for researching
factors and causes contributing to accidental events. The event tree method
focuses on events that occur after some critical event, such as “loss of power
(black out)” while the fault tree method examines all events leading up to the
critical event. The event tree is an ‘inductive’ type of analysis, while the fault tree
is a ‘deductive’ type of analysis. Event tree analyses are helpful for analysing
mitigating measures that can help reduce the consequences of some critical
occurrence. Fault tree analyses are concerned with investigating underlying causes
that result in accidental events such as “loss of power”. Bayesian network analysis
involves constructing a graphical model that shows the probabilistic
interdependencies between a set of variables. In the marine industry, Bayesian
networks can be used for decision support for maintenance planning and riskrelated issues (Friis-Hansen, 2000). Within the Safeship project, a Bayesian net
was developed for calculation of collision probabilities of ships with wind farms
(Germanischer Lloyd et al., 2005). This was constructed to help serve as the basis
for assessing the risk reduction possibilities of AIS and VTM.
16
SSPA REPORT NO:
2005 4028
AUTHORS:
Review of Current Methods/Guidelines and State of the
Art
2
REVIEW OF CURRENT METHODS/GUIDELINES
AND STATE OF THE ART
2.1
National Guidelines
This chapter presents a review of guidelines in use in other jurisdictions and
methodologies recently used to conduct navigational risk assessments for offshore
wind farms. All information is based on publications or personal contact with
individuals involved with guidelines or projects. The review includes a summary
of the guidelines used or process followed in all countries where offshore wind
farms have been constructed as of 2007 (except Japan, which has one small
offshore installation consisting of 2 individual turbines installed inside a
breakwater). A number of other countries where offshore wind farms are in the
planning stages have also been included, particularly those countries where
navigational risk assessments have been carried out as part of the planning
process.
2.1.1
Belgium
Belgium currently has one offshore wind farm under construction - Thornton
Bank. The first phase of this wind farm is expected to become operational in
2008. A risk assessment was performed by Germanischer Lloyd (GL) (Neuhaus
and Thrun, 2003) as part of the approvals process. This assessment followed the
GL guidelines published in 2002 (see section 2.1.5 for a description of the GL
guidelines). There are currently two new projects planned within the near future
for Belgium, and the risk assessments will be based on the state-of-the-art of other
risk assessments in Europe (Di Marcantonio, 2007).
2.1.2
Canada
In Canada, there are currently no offshore wind farms that have been constructed
or in operation. There are, however, a substantial number of wind farm
installations on land and on shorelines at a variety of locations across Canada.
There is currently an offshore wind development in the planning stages in British
Columbia, on the west coast of Canada. The NaiKun wind development is
currently in the process of conducting an environmental assessment review, which
17
SSPA REPORT NO:
2005 4028
AUTHORS:
Art
will include information on navigation and marine safety (Pottinger Gaherty,
2007). There are no specific Canadian guidelines for conducting a navigational
risk assessment.
2.1.3
Denmark
Denmark currently (in 2007) has 8 offshore wind farms in operation (Dansk
Vindmølleforening). The Danish Energy Authority is responsible for granting
approval for offshore wind energy projects in Denmark and for deciding whether
a specific project requires an environmental impact assessment (EIA)
(BalticMaster 2007). The Danish procedures for granting approval for offshore
wind farms have developed over time as experience has been gained with offshore
wind farms. The Danish Energy Authority screens potential wind farm locations
at sea to assess suitability of locations. The screening process includes
consultation with the general public and concerned authorities with regards to
environmental conditions and concerns, navigational safety, and
aesthetic/landscape related concerns. The tender process incorporates results of
the screening, through requirements for tenderers for the EIA process and for
specific location and design requirements (Danish Energy Authority 2005).
The navigational risk assessment carried out for Rødsand II, a large offshore wind
energy development planned for 2010, is an example of the type of navigational
safety assessment currently being carried out for developments in Danish waters.
The proposed Rødsand II project consists of 92 turbines plus 3 possible test
turbines (Christensen 2007). The risk assessment followed the steps in the
International Maritime Organization’s (IMO’s) Formal Safety Assessment
process. A hazard identification procedure was carried out to identify the risks.
Collision frequencies were evaluated for a number of different scenarios.
Automatic Identification Systems (AIS) data for the area was used to identify
shipping lanes, ship types, distributions, etc. (Christensen 2007). Ship traffic for
the year 2020 was forecast, and scenarios were considered for both current traffic
and the traffic levels for 2020. Both powered and drifting collision frequencies
were estimated using DNV’s MARCS (Marine Accident Risk Calculation
System) model. DNV’s model is discussed in more detail in Chapter 3.
2.1.4
France
Construction of the Côte d’Albâtre offshore wind farm, the first offshore wind
farm in France, is expected to begin in 2008 (Enertrag 2007). The wind energy
development will be located 6 to 11 km off the coast of Normandy, and will
consist of 21 turbines of 5 MW capacity each. The Côte d’Albâtre development is
the first offshore wind farm to be constructed in France, and was the only one
approved during the first French offshore wind energy tender (closed in 2005).
French government maritime agencies were involved the consultation and
18
SSPA REPORT NO:
2005 4028
AUTHORS:
Art
approvals process. Potential effects on other users of the sea were considered as
part of the tender evaluation process.
2.1.5
Germany
Germany has considerable experience with planning offshore wind power
generation, although to date (2007) only 2 offshore wind installations have been
built, each with only 1 turbine. The procedures for application and approval are
well established and include an assessment of the navigational risks. Detailed
guidelines on how to carry out and present risk assessment studies have been
published by Germanischer Lloyd (GL).
Currently thirteen projects in the North Sea and two projects in the Baltic have
been approved for construction. Almost twenty additional wind farms are planned
in the North Sea. A further four projects are planned in the Baltic and two wind
farms did not get approval because of environmental concerns. New wind farms
have to be approved by the German authorities (Bundesamt für Seeschifffahrt und
Hydrography). For the risk analysis Germanischer Lloyd’s guidelines are to be
followed (Richtlinie zur Erstellung von technischen Risikoanalysen für OffshoreWindparks, 2002).
2.1.5.1
Germanischer Lloyd Guidelines
The risk analysis is carried out using the following base data:
•
•
•
•
•
•
•
description of the planned wind farm including position, dimensions,
number and arrangement of the wind energy plants, substation, cable,
operation and safety concept
detailed description of the individual wind energy plants (construction,
materials, etc.) and related auxiliary devices
the sea area including the meteorological data
the maritime traffic including fishery
other offshore installations
the air traffic
the coastal protection equipment/procedures including salvage and
rescue
It is recommended that a risk analysis be carried out for each phase of the project
(installation, operation and removal), and a risk analysis is mandatory for the
operational phase. The following assumptions are made for the analysis:
•
•
future ship techniques and ship traffic are not included in the analysis
negligent actions, failures, omissions and mistakes are disregarded
19
SSPA REPORT NO:
2005 4028
AUTHORS:
Art
•
•
•
•
•
•
•
•
warlike and criminal actions are ignored
aircraft accidents are not quantified
the wind energy plant is assumed to be inherently safe and analysis of
function and stability of the plants are not included in the risk analysis
small vessels (<500 t) are included qualitatively
possible extensions of the plants or substations are not part of the
analysis
ship-ship collisions within the wind farm are not included
no calculations nor simulations of spills after a collision are done and
their consequences are not part of the analysis
maintenance works during operation are disregarded
For each scenario considered as part of the analysis, probability and consequences
should be determined. The overall risk should then be compared to published
statistical values for the risk and the results evaluated. The risk assessments
should be done both qualitatively, where the occurrence probability and the
consequences of the identified risks are described in subjective terms, and
quantitatively, where calculations/numerical estimates are obtained for both
probability and consequences.
According to the GL guidelines, the following analytical methods can be used:
•
•
•
•
•
•
qualitative, formal hazard analysis: this is a deterministic, formal and
inductive method for the identification of hazards and can be used as
the base for fault tree and the Monte-Carlo analysis. All systems have
to be included in the simulation and the fault or undesirable event is
identified. In addition, the consequences have to be considered. The
severity of the incident and the probability are estimated and the risk
priority number is calculated.
risk matrix: a risk priority number is estimated for each scenario
identified and placed in a risk matrix. Values between one and three
indicate a low risk, while four is seen as critical, and all values from
four to seven are considered unacceptable and must be analysed in
depth using quantitative methods.
“Pedersen” method: this method can be used for the scenario “collision
maneuverable ship – wind energy plant”. A Gaussian distribution is
assumed for the shipping traffic without restrictions and an
unsymmetrical distribution for bouyed fairways.
fault tree analysis: this method can be used for the scenarios “collision
maneuverable ship – wind energy plant” and “collision disabled ship –
wind energy plant”. As a minimum, this should be done graphically.
Monte-Carlo simulation: this method can be used for the “collision
disabled ship – wind energy plant”.
Consequence analysis: Potential spills of hazardous materials from the
damaged ship and the wind energy plant need to be considered for all
collision cases. Oil spill (both fuel and cargo) is of particular concern.
20
SSPA REPORT NO:
2005 4028
AUTHORS:
Art
For each ship type and size, probability of an oil spill and mean
amount of oil spilled after a collision should be estimated.
GL will publish new guidelines in the near future based on new experience
gained, new data, and new models.
2.1.6
Ireland
A navigational risk assessment is required for any proposed offshore wind park in
Ireland according to the Irish authorities. Initially, a Foreshore Licence is required
to allow investigations to access the suitability of the site. Where the site is
deemed suitable, an application may be made for a Foreshore Lease to construct
an Offshore Electricity Generating Station. It should be noted that maritime safety
would be a primary concern when assessing a site’s suitability. For this reason a
number of statutory bodies are consulted, including Irish Lights (statutory body
for Irish Lighthouses), Irish Aviation Authority and the Marine Safety Directorate.
In addition, a member of the Marine Safety Directorate sits on the Marine
Licensing Vetting Committee (MLVC) which advises the Minister of
Communications, Marine and Natural Resources on whether a Foreshore Licence
/ Lease should be granted to an applicant.
Although the risk assessments are generic in nature the developer is required to
address the specifics associated with the particular proposed development to
determine the degree of impact on the safety of navigation. The following should
be addressed:
•
•
•
The proximity of the wind park to main shipping routes.
The proximity of the park to shipping lanes, traffic separation schemes,
port entry channels, navigation marks, etc. The above are generic and
require a study of shipping activity - commercial transit traffic, regular
ferry routes, fishing and leisure craft associated with the area. This
information can be sourced from local harbour authorities, fishing co-ops
and yacht / sailing clubs.
The specifics of the footprint of the proposed wind park should be given
careful consideration in that it may result in radar interference and visual
interference where one vessel may be obscured from another vessel
because, for example, the wind park was arranged as a block.
Guidelines similar to the ones published in the U.K. on navigational risk
assessments should be used in Ireland, but common sense should prevail and as a
minimum the developer should engage the services of a marine consultant who
would have a full understanding of the requirements from a navigational safety
perspective (Foley, 2007).
21
SSPA REPORT NO:
2005 4028
AUTHORS:
Art
2.1.7
The Netherlands
In the Netherlands, the navigational risk assessment has a strong focus on
consequences, and modelling to determine effects of a potential oil spill is
required. The navigational risk assessment carried out for the Egmond aan Zee
offshore wind farm can serve as an example of what is currently required by the
authorities in the Netherlands. This wind farm, a development with 36 turbines
located approximately 10 km off the Dutch coast, commenced operations in 2007.
The navigational risks assessment carried out for Egmond aan Zee consisted of
the following main elements:
• assessment of the collision probabilities due to the presence of the wind
farm
• assessment of impacts resulting from oil and chemical spills due to
shipping collisions with the wind farm
• assessment of the effects of the wind farm on shipping radar.
(Kleissen, 2006).
The SAMSON model was used to estimate collision probabilities (see Section
3.4.1 for a discussion of this model). The probability of a passing ship ramming or
drifting against a wind turbine was estimated. Consequence modelling was also
carried out to estimate the damage to ships and wind turbine structures. Human
consequences from ramming and drifting incidents were estimated. Hypothetical
spills were also modelled, to determine the potential effects of any spills on the
coastline. Effects of the wind farm on shipping radar were studied using a full
mission bridge simulator (see Section 3.5 for a brief summary). Furthermore, an
estimation of the effects of the presence of the wind farm on shipping outside the
location of the wind farm was made (Kleissen, 2006).
2.1.8
Norway
The Havsul project is the first offshore wind energy project in Norway to apply
for a license from the NVE, the Norwegian regulatory body for energy. According
to email contact with the author of the risk analysis for this project, the assessment
has been based on purely nautical problems. A risk analysis for collisions has not
been performed, because there was not a sufficient data basis for a scientifically
credible risk estimation. Calculations would therefore have a lack of statistical
significance. Moreover the wind farm is located in a shallow fairway area which
is not trafficked by large ships.
22
SSPA REPORT NO:
2005 4028
AUTHORS:
Art
Figure 2.1. Full mission simulations were used for navigational farm analyses in Norway
(photos and illustrations provided by Norvald Kjerstad, Professor at Høgskolen i Ålesund).
The study for the Havsul project was mainly based on simulations with a focus on
navigational systems, anchoring and rescue issues. Simulations were carried out
for a ship sailing close to the farm and to determine if the wind power plants could
be used as navigational lights. Experts such as pilots, captains and mates were
invited to navigate virtually through the wind farm. The conclusion was that the
positive effects of the wind farm make up for the negative impacts (Dirdal (2007)
and Kjerstad (2005, 2006, and 2007)).
2.1.9
Spain
A new Spanish law allowing the construction of offshore wind parks came in to
force on August 1, 2007 (Burgermeister, 2007). This new law simplifies the
authorisation process by giving the power to just one office (European Wind
Energy Association, 2007), and it is expected to pave the way for the construction
of offshore wind farms in Spain. Although Spain is the world’s second leading
producer of wind power (Burgermeister, 2007) there are currently no offshore
wind farms.
The Ministry of Industry in Spain is carrying out a study to identify the best sites
along the Spanish coast for offshore wind farms, which is expected to be
completed in July 2008. A programme is also being launched to establish a
licensing procedure for Spain (European Wind Energy Association, 2007).
2.1.10
Sweden
So far there are no national guidelines, official policies or governing documents
with respect to navigational risk assessment for offshore wind farms. This report
23
SSPA REPORT NO:
2005 4028
AUTHORS:
Art
is, however, intended to provide input for such a document. The National Board
of Housing, Building and Planning (Boverket), however, has published a book on
the planning and approval processes for wind power plants (Boverket 2003). A
new edited version is under way. A brief description of the Swedish permission
process is provided in the BalticMaster project’s case study of Kriegers Flak
(BalticMaster 2007).
In Sweden, SSPA has performed a number of risk analysis studies on the
navigational risks associated with offshore wind farms. Figure 2.3 shows the steps
carried out and the components included in the analyses performed by SSPA. The
focus of the studies has been on ship traffic but fishing vessels and pleasure craft
have also been included to a certain extent. A list of some of the risk analyses
performed by SSPA is presented below and the location of the proposed offshore
wind farms is illustrated in Figure 2.2.
•
Storgrundet, off the Coast of
Söderhamns region, in the
southern Gulf of Bothnia
(Johansson et al 2007b).
•
Finngrunden in the southern
Gulf of Bothnia, off the coast of
Gävle, Sweden (Johansson et al
2007a).
•
Utgrunden II in the southern
Kalmar sound (Baltic Sea)
(Sandkvist och Hammar 2002).
•
Hanöbukten (Johansson och
Forsman 2007).
•
Kriegers flak II in the southern
Baltic Sea (within the Swedish
exclusive economic zone).
(Hammar och van Berlekom
2004).
•
Skottarevet in the Kattegatt,
off shore from Falkenberg
(Johansson et al 2005 and
Forsman et al 2007).
•
Fladen in the Kattegatt
(Magnusson 2002).
Figure 2.2. Map showing locations of proposed
offshore wind farms for which SSPA has carried
out navigational risk assessments.
24
SSPA REPORT NO:
2005 4028
AUTHORS:
Art
Introduction
Background
Purpose
Delimitations
Method
Sea traffic
AIS-data, VMS,
port statistics
Hazard Identification
Climatological and
environmental data
Risk Analysis
Qualitative and Quantitative
Wind, waves, current, ice,
fog, seabed composition,
depth
Consequence
Probability
Life, environment, material,
function
Powered collision
Drifting collision
Evaluation of the risk
Puts results in a context
Risk reduction measures
Examples for reducing probability and
consequences, respectively.
Uncertainty/Sensitivity
Accident Preparedness
Safety measures and control programs
Figure 2.3. SSPA’s procedure for assessing navigational risks associated with offshore wind
farms.
25
SSPA REPORT NO:
2005 4028
AUTHORS:
Art
2.1.11
United Kingdom
The UK’s Department of Trade and Industry (DTI) (now known as Department
for Business, Enterprise, and Regulatory Reform (BERR)) have published a
methodology to guide offshore wind farm developers in assessing marine
navigational safety risks of their proposed wind farms (DTI, 2005). The
methodology states that developers should base their submissions on a Formal
Safety Assessment, and should use “numerical modelling and / or other
techniques and tools of assessment acceptable to government and capable of
producing results that are also acceptable to government”. This allows developers
to select tools and methods that are appropriate to the site under consideration,
rather than prescribing specific methods to be used by all. The methodology was
produced in association with the Maritime Coastguard Agency (MCA) and BMT
Renewables Limited. BMT Renewables also participated in the Safety at Sea
project and the harmonised methods recommended in that project are in line with
DTI’s methodology. The Maritime and Coastguard Agency (MCA) provides
guidance and recommendations on navigational safety issues in their document
“Marine Guidance Note 275” (MCA 2004), which should be used in conjunction
with the document “Methodology for Assessing the Marine Navigational Safety
Risks of Offshore Wind Farms” (DTI 2005). The marine guidance note will be
updated in 2008, and proposed updates include guidance for emergency response
issues and an annex with an MCA shipping template for assessing wind farm
boundary distances from shipping routes.
The key features of DTI’s methodology are stated in DTI (2005) as follows:
- Define a Scope & Depth of the submission proportionate to the
scale of the development and the magnitude of the risks
- Estimate “base case” level of risk
- Predict “future case” level of risk
- Create a hazard log
- Define risk controls and create a risk control log
- Predict “base case with wind farm” level of risk
- Predict “future case with wind farm” level of risk
- Submission
(DTI, 2005)
The Marine Guidance Note on navigational safety (MCA 2004) addresses issues
such as site position (including guidance on traffic surveys), structures and safety
zones (covered in Annex 1); “developments, navigation, collision avoidance and
communications (Annex 2), safety and mitigation measures recommended for
OREI during construction, operation and decommissioning (Annex 3), search and
rescue matters (Annex 4)” (MCA 2004).
All offshore wind farm projects in the UK must first obtain a licence from the
Crown Estates, which owns the seabed of the UK out to the 12 nautical mile
territorial limit. The Crown Estates has awarded agreements for leases in two
“rounds”. The first round of agreements was granted in 2001, and the leases are
26
SSPA REPORT NO:
2005 4028
AUTHORS:
Art
for a period of 20 years. A total of 18 companies were awarded leases during this
round. The second round, announced in 2003, resulted in 12 companies/consortia
being awarded leases in 3 strategic areas. These leases are for a 40 year period.
Companies must be pre-qualified before they can be considered for a lease. To
pre-qualify, they must have financial standing, offshore development expertise,
and wind turbine expertise.
Once a developer has obtained a lease for an offshore wind farm site, they must
obtain a number of statutory consents, and must publicise an application, to ensure
the public and interested organisations have an opportunity to comment and
express concerns before a decision is made. Developers are required to provide a
comprehensive assessment of likely impact on factors such as marine
environment, visual impact, fishing, and shipping. The assessment must be carried
out for all phases of the development: construction, operation, and
decommissioning (UK Department for Business, Enterprise & Regulatory
Reform, 2007). This assessment should be described in the developer’s
Environmental Impact Assessment, and included in a resulting Environmental
Statement.
There are many wind farms in operation or under development in the UK. As of
October 2007, there were six operational wind farms (British Wind Energy
Association (2007)): Barrow, Blyth Offshore, Kentish Flats, North Hoyle, Scroby
Sands, and Burbo Bank. A further six were reported to be under construction:
Beatrice, Inner Dowsing, Lynn, Rhyl Flats, and Solway Firth/Robin Rigg A and
B. At least six more projects had been approved, and a number have been
submitted for approval. To date there have been no recorded incidents for a ship
collision with a wind turbine structure (personal communication, Navigation
Safety Branch, November 2007).
Some examples of the range of techniques used for navigational risk assessment
are as follows:
•
•
Gunfleet Sands Wind Farm: This project has been approved for
development. For the collision risk assessment carried out for this project,
the COLLIDE 2.60 model was used to estimate collision frequency
(Safetec Ltd., 2002). In terms of consequence assessment, a log-log plot
of annual collision frequency versus impact energy was generated for the
two park locations which were identified to have the highest and lowest
annual collision frequencies.
Burbo Bank Offshore Wind Farm: This wind farm was officially
inaugurated in October 2007. For the navigation risk assessment,
quantitative risk modelling was carried out using Anatec’s COLLRISK
model (Anatec, 2002). Passing drifting ship and anchor drifting ship
collision rates were estimated. In addition, a fishing vessel risk assessment
was carried out.
27
SSPA REPORT NO:
2005 4028
AUTHORS:
Art
2.1.11.1
Safety Zones
The UK recently introduced new regulations regarding safety zones around or
adjacent to an offshore renewable energy installations. These regulations, “The
Electricity (Offshore Generating Stations) (Safety Zones) (Application Procedures
and Control of Access) Regulations 2007 (SI No 2007/1948)” came in to effect in
August 2007. Standard dimension of the safety zone is 500 metres during
construction (which is the maximum permissible under international law), and 50
metres during the operational phase of an installation’s life (Department for
Business Enterprise & Regulatory Reform, 2007). If it is considered that a larger
safety zone is required for a specific case, an application for consent under
Section 36 of the Electricity Act must be made. The requirement for a larger
safety zone should be considered as part of the navigational safety assessment.
2.1.12
USA
In the USA, as of June 2007, there had not yet been any offshore wind farms
constructed, although a number of projects have been proposed (Butterfield et al.,
2007). The “Cape Wind” project in Massachusetts, the first project development
in the USA (Ram, 2004), is a proposal for 130 wind turbines. The Long Island
Power Authority had proposed a project off of Long Island, New York (Ram,
2004), and Winergy is considering a number of sites along the eastern seaboard of
the USA (Winergy, 2007). Their initial project, Plum Island Wind Park, is a small
scale research, development, and demonstration project to be located off the
northeastern tip of Long Island, New York (Winergy, 2007). There is also a
project in the development stages in Texas, being developed by Galveston
Offshore Wind. The company plans to have 50 wind turbines installed by 2010
(Fowler, 2007).
There are no published US guidelines for navigational risk assessments for
offshore wind farms. For the Cape Wind Farm, however, a navigational risk
assessment was carried out and submitted to the US Army Corps of Engineers
(USACE) as part of an environmental assessment conducted in 2004. At that time,
the USACE was the lead agency for permitting offshore wind facilities, based on
Section 10 of the Rivers and Harbors Act (Ram, 2004). This states that permits are
required for any structures altering or obstructing navigable waters. The Energy
Policy Act of 2005 granted the Interior Department’s Minerals Management
Service (MMS) new responsibilities related to renewable energy and now this
national agency is the lead agency for permitting and regulatory oversight of
offshore wind energy projects sited on federal offshore lands, on the outer
continental shelf (OCS). The OCS extends from 3 nautical miles (nm) from the
coastline out to 200 nm, except for Texas and Florida, where the state jurisdiction
extends to 9 nm from the coastline (Ram, 2004). Although the MMS now has the
role of lead agency and responsibility for coordinating the permitting process, the
regulations pertaining to the USACE permits are still in place.
28
SSPA REPORT NO:
2005 4028
AUTHORS:
Art
2.2
2.2.1
International Research Programmes and
Harmonisation of National Assessment Schemes
Safeship
The project “Reduction of Ship Collision Risks for Offshore Wind Farms” (with
acronym “SAFESHIP), was a 2-year project that was completed in January 2005.
The project was co-financed by the European Commission as part of the 5th
Framework RTD Programme (den Boon et al., 2005). The overall objective of the
project was “to reduce the risks of ship collisions with offshore wind farms by
development of appropriate cost-effective technologies and risk assessment
methodologies, thereby reducing the production costs of offshore wind energy and
removing development barriers” (den Boon et al., 2005).
In terms of the modelling of collision risks, the project compared the frequency
models of Germanischer Lloyd AG (GL) and of the Maritime Research Institute
Netherlands (MARIN), as both of these organisations were partners in the project.
The model used by Det Norske Veritas (DNV) was also included in some of the
comparison work. Models for both collisions of a powered ship with a wind farm
and for collisions of a drifting ship with a wind farm were compared. After model
comparisons, changes were made to the models to result in what was hoped to be
more harmonised predictions. The collision models compared and developed in
SAFESHIP are further discussed under Section 3.4.1, Ship – Wind Farm Collision
Probability Estimation. Consequence modelling within the project was carried out
using finite element modelling (using LS-Dyna).
With respect to the work carried out on risk reduction measures and technologies,
the SAFESHIP project produced the following:
•
•
•
a catalogue of cost-effective methods and technologies; the main result is
the conclusion that AIS (Automatic Identification Systems) is the most
effective risk reducing method;
a detailed design of fendering for the HV station of a wind farm;
an Emergency Response Management Plan for the 120 MW Q7-WP wind
farm, to be used as a model for other wind farms.”
In addition, the project work resulted in the conclusion that placing fendering
around offshore wind turbines was not cost-effective, although it was technically
feasible.
29
SSPA REPORT NO:
2005 4028
AUTHORS:
Art
2.2.2
Safety at Sea project
The partner countries and organisations of the Safety at Sea research program (an
Interreg North Sea Region project) have developed a procedure of harmonised
methods for carrying out a marine navigational safety assessment. The description
is in line with the UK DTI Guidelines. The policy recommendation as part of the
output of the Safety at Sea project is “that these methods are adopted by EMSA
and developed into a standard to cover all EU waters and extended to cover all
offshore renewable energy including tidal and wave energy.” In addition, the EU
Member State Maritime Administrations are invited to apply the draft procedure
developed as part of the Safety at Sea program during the development of their
own regulatory requirements (Starling, 2007).
The methods proposed and presented, however, do not provide details for
procedures such as calculation of collision probability, and in reality there is no
guarantee that the application of this general methodology would result in
consistent results. The Safety at Sea report (BMT, 2005) states that there is a wide
range of risk assessment techniques available, and those selected for a specific
project should be in line with the scale of the project and acceptable to the
involved Maritime Administration and regulatory bodies. This implies that a
range of technical analyses techniques may be used for different parks and in
different countries.
The Safety at Sea project also included a demonstration research project on
offshore wind farm risk management (there were six demonstration projects in
total). The project as a whole was co-funded by the Interreg IIIB North Sea
Region Programme, and its primary aim is “to reduce the probability and impact
of accidents in the North Sea”. It was a three-year project that began in September
2004 (www.safetyatsea.se) and more than 20 organisations from countries
surrounding the North Sea were involved. The project was managed by the
Norwegian Coastal Administration.
The offshore wind farm risk management project resulted in a number of
deliverables. A cumulative quantified risk assessment was carried out for an
arbitrary sea area which had up to five wind farms and included a large range of
other features including islands, oil installations, ports, etc. Simulated current and
future marine traffic was created and used to carry out the risk assessment.
Further, the project resulted in risk control provisions which were identified for an
arbitrary wind farm within the arbitrary sea area (BMT Renewables, 2005).
One recommendation was that proportionality of the project be assessed to
determine the amount of detail required in the submission and the Navigation Risk
Assessment. A continuum of activities is described for support of the navigation
risk assessment, starting with “area traffic modelling/assessment of the strategic
area”, and ending with specific traffic bridge control simulations and site specific
trials that may be required for assessing risk control options and for more
complicated projects.
30
SSPA REPORT NO:
2005 4028
AUTHORS:
Art
2.2.3
Harmonisation discussions in Germany
In Germany the authorities appointed a group of experts to discuss harmonisation
of the assumptions made in risk assessments (Bundesministerium für Verkehr-,
Bau und Wohnungswesen 2005). The scope was also to find and agree on risk
acceptance criteria. The goal of the group of experts was to find generally
accepted values for the model assumptions, the collision frequency, the mean oil
spill per year, risk reducing measures, the carving depth of cables and minimum
distances. The results are seen as state-of-the-art which will be modified with
increased knowledge in the field and adjusted to future developments.
31
SSPA REPORT NO:
2005 4028
AUTHORS:
Navigational Risk Assessment Methodology
3
NAVIGATIONAL RISK ASSESSMENT
METHODOLOGY
The purpose of this chapter is to describe the different steps of a navigational risk
assessment for offshore wind farms. It includes hazard identification, estimation
of collision probability and consequences, discussion on uncertainty/sensitivity as
well as risk acceptability. Effects on radar, radio, navigation equipment etc and on
search and rescue operations are also included in this chapter but risk reduction
measures are presented separately in the next chapter.
3.1
Initial qualitative assessment
As an initial step of the navigational risk assessment for proposed or existing
offshore wind farms, qualitative hazard identification should be performed. All
parties involved in the project and relevant maritime stakeholders should be
addressed and asked for their concerns and opinions regarding possible hazards
associated with the establishment of the wind farm. This may be conducted by
public hearing, interviews or structured brainstorming sessions in selected groups
of stakeholders. As an output, a catalogue of possible potentially hazardous
scenarios can be identified.
Based on further qualitative considerations on the likelihood of the respective
scenarios to occur and the potential consequences of the scenarios, an initial
ranking and selection of prioritised scenarios that need to be further analysed
quantitatively can be identified.
Taking into consideration the park location, its size and the output of the initial
assessment, decisions are taken on the needs and levels of additional detailed
studies and a detailed plan is outlined for further risk assessment studies.
The following sections describe a general methodology for detailed assessment of
navigational risks associated with offshore wind farms.
3.2
Definition of Types of Risks to be Considered
The construction and operation of a wind farm will potentially have an effect on
the risk of many types of incidents that involve ships operating in waters in the
32
SSPA REPORT NO:
2005 4028
AUTHORS:
vicinity of the wind farm. Navigational risks which may be introduced or changed
by the establishment of an off-shore wind farm include the following:
•
•
•
•
Risk of a ship colliding with or contacting a wind turbine or wind farm
structure
Risk of ship to ship collision resulting from change in navigation to avoid
the wind farm area
Grounding Risks
Possible secondary risks resulting from effects of the wind farm on for
example radar operation
A methodology should include an assessment of each of these types of risks.
These risks will be different during different phases of the project. Although the
project construction time and decommissioning are relatively short compared to
the wind farm operational phase, they should still be given some consideration
during a risk assessment.
It is also important to select a time frame to be considered for future scenarios.
The “base case” or “pre-wind farm” risk for ship-to-ship collisions and grounding
needs to be compared to risk estimated for collisions when the wind farm is
operational.
In addition, comparisons should be made for future scenarios of increased ship
traffic and for changes to the ships such as increased average speed, and changes
to draught and tonnage.
Finally, there should be some consideration given to the changes in risk that may
result from future developments for wind farms (larger turbines, different
foundation types, etc.) and for the cumulative risk that could result from the
establishment of several wind farms along a navigation route and possible risk
interactions.
Figure 3.1 illustrates the aspects of risk changes over time and from a regional
scale perspective.
The quantitative risk estimation methods developed and analysed in this study are
basically focused on estimating absolute risk figures associated with wind farms
but, as also illustrated in the figure, for overall assessment, relative risk figures are
generally more conclusive and provide important input in the decision making
process. Risk assessments based on relative risk considerations are also less
sensitive with regard to uncertainties and assumptions in critical numerical input
parameters.
33
SSPA REPORT NO:
2005 4028
AUTHORS:
Regional
scale
National level
Navigational
safety policy
Regional
level
Cumulative effects and interactions caused by other new farms
Future expected/predicted changes in wind energy techniques
Future expected/predicted changes in ship traffic pattern/volume
Local
site
level
Present
Construction phase
Collision risk during operational
phase of the park
Decommission phase
Time
scale
situation
Absolute risk consideration
Relative risk consideration
Figure 3.1. Various risk aspects considered from regional perspective and with
respect to time.
3.3
Data Sources and Inputs to Analyses
The following sections provide a description of methods used for estimating
probability and consequences of accident scenarios associated with interactions
between ships and wind farms. Ship to ship collision and grounding are also
discussed.
To estimate the probability of collisions between ship and wind farm, the
following type of information is generally required:
•
Wind farm data including:
o Position of each wind turbine
o Distance between turbines
o Pile diameter
o Hub height
o Rotor diameter
o Installations of navigational aids (lights etc) at the wind farm
34
SSPA REPORT NO:
2005 4028
AUTHORS:
•
Vessel traffic information in the vicinity of the proposed wind farm,
including:
o Position of typical shipping lanes/operating routes (from AIS data
and other sources)
o Number of vessels (from AIS data and other sources)
o Types of vessels such as cargo, tanker, passenger, etc. (from
sources such as AIS data and general statistics for the Baltic Sea)
o Characteristics of vessels such as size, length, breadth, draught,
operating speed (from sources such as AIS data and general
statistics)
o Seasonal traffic variations (from AIS data, etc.)
o Day/night traffic variations (data sources include local ports)
o Future traffic scenarios (from Helcom, VTT, The Institute of
Shipping Analysis in Göteborg, etc.)
o Distance from the shipping lane to the wind farm (estimated from
AIS data; estimations for lanes shifted to new location)
o Standard deviation and mean for lateral distribution in crosssection (from AIS data (histogram) or general estimations)
o Statistical distribution for course deviation, e.g. standard
deviation and mean for Gaussian distribution in cross-section
(may be possible to derive from AIS data or statistics)
•
Climatological data including:
o Wind speed distribution and wind direction distribution (10
meters over sea level)
o Wave information
o Current information
o Ice conditions
o Fog conditions (to assess hazards such as reduced visibility)
•
Site data including:
o Coast line geometry
o Water depth, bathymetry and sea level variations
o Type of sea bottom such as rock, clay, sand, etc.
•
Frequency of machinery breakdown– blackout
•
Ship self repair function (time for self repair)– duration of blackout
•
Probability of unsuccessful emergency anchoring
•
Tug boat assistance information including:
o Distance from tug boat position to disabled ship
o Operating speed of the tug boat (depends on the weather
conditions)
35
SSPA REPORT NO:
2005 4028
AUTHORS:
o Time to activate the tug boat, to connect and take control of the
disabled vessel
•
Probability of:
o human failure during planning and execution of the passage of an
object (navigation past an obstacle)
o technical failure of navigational equipment or watchkeeping
failure due to causes such as lack of attention during lookout on
the bridge or bad visibility
o failure of the wind farm safety equipment/crew or a potential
stand-by boat to warn the passing ship in time to avoid a collision
o the crew onboard being unable to react in time to correct the
navigational error (dependent on the distance between ship and
wind farm)
For future vessel traffic and ship sizes, information can be obtained from sources
such as the Baltic Maritime Outlook (2006), which provides estimates on future
maritime freight flow in the Baltic Sea Region, including projected flows on
specific transport corridors. In addition, the characteristics of specific ship types
on order as provided in sources such as Lloyd’s Fairplay can give an indication
of future ship sizes (DWT, draught) and speeds.
For consequence analysis, information such as the following is required:
•
•
•
•
Information on the wind turbine structure, foundation type and distance
between blade and water surface
Information on soil types
Environmental information such as specific species using the area,
sensitive shoreline areas, etc.
Information on vessel types and characteristics, including cargo and
bunker fuel quantities for estimating collision results and potential spills
Data sources on vessel traffic information for Swedish waters include:
•
•
•
•
Swedish Maritime Administration, which can provide AIS-data
Ferry companies operating regular line services, such as Stena
Port Authorities for information on line traffic. Number of ship arrivals
in different ports is available from Sveriges Hamnar (2007).
Information about fishing boats (VMS-data) can be obtained from the
Swedish Board of Fisheries and information about recreational boats
from Gästhamnguiden (2007).
36
SSPA REPORT NO:
2005 4028
AUTHORS:
3.4
Analysis Methods
A variety of methods have been used for estimating probability and
consequences of navigational accident scenarios associated with construction
and operation of offshore wind farms. When ship traffic in the vicinity of the
wind park is very light and data on ship traffic is limited, methods have been
more qualitative. More complete modelling and statistical analysis techniques
have been used for offshore wind farms proposed in areas with significant ship
traffic and where detailed information such as AIS data is available.
Similarly, consequence analysis methods for offshore wind farms range from
qualitative to quantitative modelling, depending on availability of data and
assessed probability of incidents. It is important to have appropriate data for
model input to ensure confidence in the results.
The following sections provide a description of methods used for estimating
probability and consequences of accident scenarios associated with interactions
between ships and wind farms. Ship to ship collision and grounding are also
discussed.
3.4.1
Ship – Wind Farm Collision Probability Estimation
There are a number of different models for estimating the probability of ships
colliding with offshore platforms. The models have been developed and
presented by various organisations, as shown in Table 3.1. The table also
includes references to studies where the respective models have been described
and/or applied.
37
SSPA REPORT NO:
2005 4028
AUTHORS:
Table 3.1. Collision models, companies/organisations responsible for development, and selected
references.
Model
Company/Organisation
Selected References
COLLIDE
Safetec Nordic AS
Haugen (1998)
Spouge (1999)
Safetec (2002)
SOCRA1/SAMSON2 MARIN
van der Tak and Glansdorp (Year unknown)
(Maritime Research
van der Tak und Rudolph (2003)
Institute Netherlands)
van der Tak (2005a)
van der Tak (2005b)
SAFESHIP (2005)
SAFESHIP (2006)
Kleissen (2006)
CRASH/ MARCS3
DNV (Det Norske Veritas)
Spouge (1999)
SAFESHIP (2005)
Christensen (2007)
COLWT
GL (Germanischer Lloyd)
Germanischer Lloyd (2002)
Neuhaus and Thrun (2003)
Otto and Petersen (2003)
Povel et al. (2004)
Otto (2004)
Povel and Petersen (2004)
SAFESHIP (2005)
SAFESHIP (2006)
Povel (2006)
COLLRISK
Anatec UK Ltd
Anatec UK Limited (2002)
SAFESHIP (2006)
DYMITRI
BMT (British Maritime
Safety at Sea (2005)
Technology) Limited
1
SOCRA (Ship Offshore platform Collision Risk Assessment) is a module in MANS (Management
Analysis North Sea).
2
SAMSON (Safety Assessment Models for Shipping and Offshore in the North Sea).
3
CRASH (Computerised Risk Assessment of Shipping Hazards), MARCS (Marine Accident Risk
Calculation System).
COLLIDE was originally developed for offshore oil platforms, but is now also
used for offshore wind farms. A possible upgrading to a new version has been
discussed (Eriksen and Haugen 2006).
MARIN’s web page indicates that the SOCRA software is used for offshore oil
platforms whilst SAMSON is generally used for offshore wind farms.
GL has issued guidelines for risk analysis for offshore wind farms (see
Germanischer Lloyd (2002)). GL has estimated the collision frequencies for a
planned offshore wind farm within the German exclusive economy zone (EEZ)
of the Kriegers Flak area of the Baltic Sea (see Otto and Petersen 2003, Povel et
38
SSPA REPORT NO:
2005 4028
AUTHORS:
al. 2004, and Otto 2004). MARIN has also estimated the collision frequencies
for Kriegers Flak (see van der Tak und Rudolph (2003) and van der Tak
(2005b)).
The various models have also been compared in previous studies. The most
recent comparative study presents a comparison between COLLRISK, COLWT
and SAMSON (see SAFESHIP (2006)). In SAFESHIP (2005) the models of
MARIN, GL and DNV were compared, with the focus of the comparison on
harmonising model assumptions. These three companies were part of a group of
experts that were appointed by German authorities to discuss harmonisation of
assumptions made in risk assessments (Bundesministerium für Verkehr-, Bau
und Wohnungswesen 2005), as discussed in Section 2.2 of this report. Van der
Tak (2005a) also compares the models in order to clarify differences, harmonise
assumptions and enlarge transparency. Van der Tak focuses on the differences
between SAMSON’s calculation model for powered collision and other
corresponding models based on Gaussian distributed offset of the ships sailing
on the lanes which pass by the wind farm (see Figures 3.2 and 3.3).
SSPA’s collision frequency model (see Appendix) is developed with the most
recent available information on model assumptions etc as a basis. The structure
of the model is similar to other models used for wind farms and offshore
platforms. The derived harmonised values from the German harmonisation
process mentioned above have become available during the work with this
research project and have partly been used in the SSPA model.
Risk analyses and assessments have also been conducted by companies and
organisation such as:
•
•
COWI (Örestads Vindkraftpark AB (2000))
Rambøll (Rambøll (2000a and 2000b), Christensen et al. (Year
unknown) and Randrup-Thomsen et al. (Year unknown))
MARIN and Germanischer Lloyd are two of the major companies performing
risk analysis for collisions of ships with offshore wind farms. This is based on
the relatively high number of offshore wind farms which have been proposed
and studied in the countries where the companies are based (the Netherlands and
Germany respectively). Together with the Technical University of Denmark they
conducted a study (SAFESHIP 2005) on the collision frequency of powered and
disabled ships with offshore wind farms, in which they compared the models
used by MARIN, Germanischer Lloyd (GL) and Det Norske Veritas (DNV). If
no other references are stated, the material presented in the rest of section 3.4.1
is based upon that study in combination with the studies of van der Tak and
Rudolph (2003), van der Tak (2005b), Kleissen (2006), Otto and Petersen
(2003), Povel et al. (2004), Otto (2004), Neuhaus and Thrun (2003), Christensen
(2007) and Bundesministerium für Verkehr-, Bau und Wohnungswesen (2005).
39
SSPA REPORT NO:
2005 4028
AUTHORS:
Further analyses and discussions on the differences between the models are
found in Chapter 5, which describes the case study of Kriegers Flak carried out
as part of this study.
40
SSPA REPORT NO:
2005 4028
AUTHORS:
Figure 3.2. Illustration of general probability calculation model based on Gaussian
distribution of ship traffic along a fairway. Collision course times causation
factor. Source: van der Tak (2005a).
Figure 3.3. MARIN’s model: Collision opportunity times Navigational Error Rate
(NER). Source: van der Tak (2005a).
41
SSPA REPORT NO:
2005 4028
AUTHORS:
3.4.1.1
Powered collision
The models used by MARIN, GL and DNV are basically quite similar. GL’s and
DNV’s models estimate the number of collision candidates and multiply this with
a causation factor, while MARIN’s model estimates the number of ramming
(collision) opportunities and multiplies this by the Navigational Error Rate (NER).
The models differ from each other in the assumptions made for the determination
of the collision candidates and the ramming opportunities.
When estimating the probability of powered collisions, the assumptions made for
the traffic around the wind farm are very important, specifically the parameters of
the lateral distribution of the ship traffic and of the centre line of the ship traffic.
These parameters are strongly dependent on the fairway (e.g. open sea, Traffic
Separation Scheme (TSS), shallow water, buoyed fairway, etc.). All models are
based on a Gaussian distribution to represent the lateral traffic distribution on the
shipping lanes. GL and DNV add a uniform distribution of 2% to the Gaussian
distribution (the width of uniform distribution is assumed to be 6 times the
standard deviation), most likely in order to represent the traffic not following any
route. MARIN calculates the non-route-bounded traffic separately.
GL uses the following parameters for the standard deviations if there are no local
factors that otherwise influence the distribution.
Table 3.2. Standard deviations according to GL (SAFESHIP 2005).
Description
Standard deviation for Gaussian
distribution [nm]
Port approach
0.2 to 0.3
Conspicuous navigational points, e.g.
0.3 to 0.4
navigational marks, buoys
Navigational channel with traffic
0.5
separation
Waypoints in wider shipping lanes
0.5 to 1.0
Waypoints in open sea areas
2.0
In order to derive the lateral distribution of the lanes on the North Sea, MARIN
has made observations of actual traffic with a partition into lanes with Traffic
Separation Scheme (TSS), connection between two schemes and with completely
free shipping lanes. In the TSS lanes the observed type of distribution is used
(which is dependent on the width of the shipping lane). For completely
unrestrained lanes a Gaussian distribution with a standard deviation of 1 nm is
used.
DNV and GL have agreed to use 1.2 times the ship breadth plus the dimension of
the object perpendicular to the sailing direction as collision width (0.2 extra ship
breadth includes an average drifting angle of 2 degrees. The kinetic energy for this
42
SSPA REPORT NO:
2005 4028
AUTHORS:
0.2 extra is assumed to cause less serious damage after a collision), while MARIN
uses only the ship breadth.
DNV and GL both use a causation factor of 3.0E-4 for a ship not taking corrective
action, which is meant to include all causes resulting from either human error or
technical failure. MARIN used the same assumption in their original model but
moved away from this because of the fact that the collision probability is very
sensitive to the distance between the centre line of the traffic lane and the object.
The collision probability is also dependent on the tail of the lateral distribution of
the ships using the lane. MARIN uses a value called NER instead. The table
below shows the relationship for the different Navigational Error Rates (NER)
according to MARIN with an offshore platform as a basis. See also Figure 3.3.
Table 3.3. Relationship for the different Navigational Error Rates (NER) according to MARIN with an
offshore platform used as a basis for comparison (data from SAFESHIP (2005)).
Type of obstacle
Single offshore
Collision/ stranding with
Offshore
platform
an island
wind turbine
Relationship for the different
Navigational Error Rates (NER)
1
6
2.5
MARIN uses the following formulas to calculate the number of powered
collisions:
x2
Pψ = ∫ e
−a
d nψ ( x )
Li
dx
x1
x2
ROk = ∑ ∑∑ pn pnψ N ik ⋅ ∫ e
n
nψ
i
−a
d nψ ( x )
Li
dx
x1
N ram min g = NER ∑ ROk
k
•
•
•
•
•
•
•
•
•
RO: Ramming opportunities
Nramming: Number of collisions
Pψ: Probability that a navigational error leads to a collision with an object
in direction ψ
Li: Ship length (in nautical miles) of a ship from ship size class i
x: Position of a ship on a shipping lane. The integration borders x1 and x2
follow from the reachability of the WPP (Wind Power Plant) from the
shipping lane. These values are dependent on the size of the WPP and the
ship size class and type.
dnψ: Distance between a point x on a part of the route and the collision
point (WPP) in direction nψ
a: Dimensionless constant for the probability of having no collision
avoidance measures taken in time after a change in course
pn: Probability of a certain load condition n
pnψ: Probability of a certain course
43
SSPA REPORT NO:
2005 4028
AUTHORS:
•
•
•
Nik: Number of ships of ship class i on the shipping lane k
i: Ship size class
NER: Navigational Error Rate
NER is an empirical value based on accident statistics and geometrical collision
candidates. No information has been found in the literature regarding the values of
NER or the constant “a” in the equations described above.
DNV uses the following formula to calculate the ship-turbine collision frequency:
c + 0.5 ( D +Wship )
FHuman = N ⋅ PHuman ⋅
∫ f ( y)dy
c −0.5 ( D +Wship )
•
•
•
•
•
•
N is the number of ships on the lane for the specific ship class per year.
PHuman is the probability of human failure (3.0·10-4).
f(y) is the transverse distribution of the ships on the navigation lane. This
is assumed to be a uniform plus a Gaussian distribution with a mean value
corresponding to the navigation lane centre-line and a standard deviation
depending on the ship type and distance to shore, shallow water or a wind
farm.
c is the distance from the turbine perpendicular to the navigation lane.
D is the turbine foundation diameter.
Wship is the width of the ship, increased by a factor of 1.2 as stated above,
for the ship class under evaluation.
GL and MARIN use different ship types and ship classes, therefore results are
difficult to compare. In the EU funded project SAFESHIP 2005 the two
companies compared their models in a sensitivity study, which showed that the
GL calculations are very sensitive to the location of the centreline and the
standard deviation of the lateral distribution of the traffic lanes. In their example
the calculations were made for a wind farm project in the North Sea outside the
Netherlands. One nm is used as the distance from the wind farm to the centre line
of the closest lanes. For details see the tables below.
44
SSPA REPORT NO:
2005 4028
AUTHORS:
Table 3.4. Probability of a ship ramming a wind farm (wind farm P12, the
Netherlands) (SAFESHIP 2005).
Model
The centre line of all traffic
All standard deviations are multiplied
used
lanes is moved with … nm
by
away from the wind farm
GL
0 nm1)
0.5 nm
1.0 nm
1.001)
0.75
0.5
0.25
0.1418
0.0481
0.0137
0.1418
0.0542
0.0049
4.3E-08
MARIN 0.0060 0.0024 0.0009 0.0060 0.0040 0.0027 0.0019
1)
These are the base cases when the centre lines positions and the standard
deviations are not changed.
Table 3.5. Sensitivity factors (SAFESHIP 2005).
Model
The centre line of all traffic
All standard deviations are multiplied
used
lanes is moved with … nm
by
away from the wind farm
0 nm1)
0.5 nm
1.0 nm
1.001)
0.75
0.5
0.25
GL
1.0
0.339
0.096
1.0
0.382
0.035
3.0E-07
MARIN 1.0
0.404
0.154
1.0
0.670
0.451
0.313
1)
These are the base cases when the centre lines positions and the standard
deviations are not changed.
Methods such as Bayesian net, event trees, statistics and published values are used
to estimate and verify causation factor. The figure below (Figure 3.4) presents a
general event tree diagram which can be used for sensitivity analysis of the
causation factor. The figure also includes a part of the tree in bigger size. The
effect of risk reducing measures on the causation factor can also be evaluated with
event trees or Bayesian net. Such studies have been conducted to evaluate the
effect of the introduction of AIS (Lützen and Friis Hansen, Year unknown) and of
different bridge designs.
45
SSPA REPORT NO:
2005 4028
AUTHORS:
46
SSPA REPORT NO:
2005 4028
AUTHORS:
Figure 3.4. Upper part: General event tree diagram Lower part: Part of event tree diagram.
GL has developed a Bayesian net to reduce the uncertainties connected with the
causation factor and to respond to the new and unknown influence of Automatic
Identification System (AIS) and Vessel Traffic Monitoring (VTM) on the
causation factor. The net is based on a Bayesian net developed by the Technical
University of Denmark for ship-ship collisions. A part of the input parameters and
their influence on the results are presented below.
The following assumptions have been made: visibility is 30 km during the day
and 20 km at night; ships have a speed of 15 knots in high visibility and a speed of
7.5 knots in poor visibility (<1 nm); and the presence of the wind farms is known
to 95% of the crews on board the ships and the officers on the watch are aware of
the presence. The alertness of the officers, however, depends on their stress level.
The alertness depends on factors that draw the attention of the officer to the wind
farm as well as the situation on the bridge.
47
SSPA REPORT NO:
2005 4028
AUTHORS:
Table 3.6. Contribution of performed “Officer on watch” tasks (SAFESHIP
2005).
Alertness
Wind farm unknown
Wind farm known
Stress level
Low
Medium High
Low
Medium High
Observing
0.55
0.5
0.458
0.7
0.65
0.6
surroundings
Observing
0.2
0.182
0.167
0.25
0.25
0.2
Radar/ AIS
Performing
0.25
0.318
0.375
0.05
0.05
0.2
other duties
The results of the calculations are shown in the table below. The causation factor
assumed by GL is 3.25E-04 (95% aware of the existence and position of the wind
farm).
Table 3.7. Influence of the assumed level of alertness
on the causation factor (SAFESHIP 2005).
Alertness
Causation factor
50%
5.97E-04
60%
5.36E-04
70%
4.76E-04
80%
4.15E-04
90%
3.55E-04
95%
3.25E-04
100%
2.95E-04
3.4.1.2
Drifting collision
The models for drifting collisions used by MARIN, GL and DNV are very much
alike, while the assumption made for some important factors like drift speed,
emergency anchoring, etc. differ. Quite a lot of assumptions have been made,
which implies a lot of uncertainties. The first assumption made is for the
probability that an engine failure occurs. The vessel starts drifting – if no
redundant propulsion is installed – with a velocity that is based on the wind and
waves, the current and ship characteristics such as ship size or loading conditions.
To stop the drift, there are possibilities such as repairing the engine failure in a
certain time or carrying out successful emergency anchoring procedures. To repair
the vessel a specific amount of time is needed, depending on the type of failure.
Emergency anchoring is only successful if the vessel is not drifting too fast, and
there are other parameters which are important such as the seabed composition,
the size of the vessel, etc. The drift can also be stopped by a salvage tug, if the tug
can reach the vessel before a collision occurs.
48
SSPA REPORT NO:
2005 4028
AUTHORS:
Similar to the powered collision models, it is possible to identify an important
parameter which results in a difference in the estimated collision frequency. A
higher drift velocity means decreased time required for the drifting ship to reach
the wind farm, leading to higher probability of failure to repair the engine in time.
It also means increased probability of failure to anchor and decreased probability
to assist the ship by a salvage tug because the time for a salvage tug to reach the
disabled vessel has decreased. The models of GL, MARIN and DNV are difficult
to compare with respect to drift velocity, because they are based on a different
breakdown in classes of vessels and different assumptions for the wave
component in the drift velocity. However, in the SAFESHIP 2005-project it was
stated that the drift velocities calculated by GL are lower than the drift velocities
used by MARIN, which explains why there are considerable differences in the
collision frequencies. The main reason for the differences in the drift velocity is
stated to be that the wave component of GL is nearly negligible, while in
MARIN’s model the wave component for the smaller Beaufort classes is even
higher than the wind component.
The assumption made for the disabled ships are however the same:
•
•
•
•
•
•
the wind and waves act in the same direction
the wind direction and velocity are kept constant during the drifting
mass effects are not included in the model
the ship moves purely in the lateral direction which is equal to one degree
of freedom
the forces consists of the wind force, the averaged second order wave force
and the resistance of the ship through the water
The effective collision width is the ship length plus the dimension of the
object perpendicular to the drifting path
MARIN uses the following formulas:
FWind =
1
ρ air ALin C dWind vb2
2
49
SSPA REPORT NO:
2005 4028
AUTHORS:
FRe sis tan ce =
FWave =
1
2
ρ water Li Tin C d v drw
_ bin
2
1
ρ water gζ b2 Li R 2
16
v drw _ bin
=
drift condition of ship i in loading condition n by wind
and waves belonging to Beaufort class b as a result of
the wind
vb
=
wind velocity for Beaufort class b
v wi
=
wind velocity
vs
=
ship speed
vT ,Wd
=
velocity of tidal current and the current induced by
drift
ρ air
=
density of air (GL=1.3 [kg/m3])
ρ water
=
density of water (GL=1024 [kg/m3])
ALin
=
lateral wind surface of ship i in loading condition n
(GL 90° relative wind direction)
Tin
=
draught of the ship i in loading condition n
Li
=
length of the ship i
ζb
=
significant wave height assumed to be generated by
Beaufort class b
ζb
=
significant wave amplitude (Hs/s)
CdWind
=
lateral wind resistance coefficient
Cd
=
lateral resistance coefficient of the underwater body of
the ship (MARIN~0.9; GL~0.855)
C dWe
=
wave induced drift coefficient (0.5)
Cdcurr
=
=
=
=
force coefficient (0.6)
wave drift coefficient
gravity
water depth
=
=
mean wave period for a given Beaufort class
displacement of the ship
R
g
h
Tp
∇
The wave drift coefficient is estimated from values from experiments and the
relation for R is dependant on the wave number k and the draft T given by:
a * (Tk ) 3 + b * (Tk ) 2 + c * (Tk ) = −1.4736 * (Tk ) 3 + 2.4765 * (Tk ) 2 − 0,0315 * (Tk )
50
SSPA REPORT NO:
2005 4028
AUTHORS:
The wave number is derived from wave theory for shallow and deep water (h ∞ : tanh(kh) 1) waves:
 2π

T
 p
2

 = kg tanh(kh)


The equilibrium between all forces is assumed and this results in a drift speed of:
v drift ,Wind (i, b, n) =
ρ air ALin C dWind 2 1 ζ b2 g R 2
vb +
8 Tin C d
ρ water Li Tin C d
GL uses the following formulas:
1
ρ air ALin C dWind (vWi + v s ) 2
2
1
FCurrent = ρ water Li Tin C dcurr (vT ,Wd + v s ) 2
2
1
FWave = ρ water g∇1 / 3ζ a2 C dWe
2
FWind =
DNV uses the following formula for the contribution of a piece of length di of the
navigation lane:
Fdrift = N ⋅ f fop ⋅
•
•
•
•
•
•
•
•
di Θi
Wdrift ∑ Pwindclass ,i (1 − Prepair ,i )(1 − Panchor )
vship 360
windclass
N is the number of ships on the lane for the specific ship class per year.
PAnchor is the probability that the ship will drop anchor and successfully
stop. This probability depends on the weather conditions and the seabed
conditions.
ffop is the frequency for failure of the propulsion machinery per hour. This
frequency depends on the ship type.
d is the length of the considered part of the navigation lane.
vship is the velocity of the cruising ship.
θ is the angle space (sector) where the sideways drifting ship will collide
with the turbine. At the turbine this corresponds to a length equal to the
ship length plus the turbine foundation diameter. This is conservative as
ships do not generally drift completely sideways.
Wdrift is the probability for the specific drift direction relative to a uniform
drift direction.
Pwind_class is the probability for the given wind class.
51
SSPA REPORT NO:
2005 4028
AUTHORS:
•
Prepair is the probability that the ship will repair the propulsion machinery
and stop drifting. This is a function of the drift duration (distance and drift
velocity). The drift velocity is assumed to be between 0.9 to 3.6 knots
depending on the wind conditions, which may be conservative.
Again, a force equilibrium is assumed to derive the drifting speed.
The tidal current is included in GL’s Monte Carlo calculations with a random
value from the current distribution for each start position. MARIN includes a
projected speed in the direction of the drift velocity, which is added to the drift
velocity. This seems to be unnecessary, because the tidal currents should result
statistically in the same collision frequency as without tides included.
Table 3.8 shows the engine failure rates used in the different models. The data
from Lloyd’s Register Fairplay (LRF) contains only engine failures which have
lead to serious delay or cases where the vessel had to be towed. The different
consulting companies agreed on the harmonised values but will continue to use
their own models with different ship types and ship sizes.
To verify the assumptions made by MARIN, GL and DNV, the Technical
University of Denmark performed a study on the probability of the propulsion
machinery to fail (SAFESHIP 2005). Two Bayesian networks were used to model
the failures of the main engine and the steering gear. The result was found to be
0.023 (1/year) for the main engine failure rate and 0.0147 (1/year) for the steering
gear failure rate.
Table 3.8. Engine failure rates used in the different models (data from SAFESHIP (2005)).
Harmo
Model
DNV
GL
MARIN
nised
1)
Based on
one propulsion more than one
LRF
Dutch
machine
propulsion
coast
machine
guard
-4
-5
-4
-5
Breakdown 4.6·10 (large) 1.34·10
2.0·10
2.9·10
4.0·10-5 2.5 10-4
-4
frequencies 2.8·10 (small
(1/ hour
(1/hour)
ships)
at sea)
Average
12 knots
16 knots
cruising
speeds
GL’s “self repair function” is based on a study made for Prince William Sound
(DNV et al 1996). The function is described in the table below and illustrated in
the Kriegers Flak case study in Chapter 5.
52
SSPA REPORT NO:
2005 4028
AUTHORS:
Table 3.9. GL’s self repair function (data from Otto und Petersen 2003).
Time after engine failure
Failure of self repair
0-10 minutes
100%
10 minutes – 1 hour
decreases from 100% to 45%
1-11 hours
11-24 hours
24 hours – 1)
1%
1)
Otto und Petersen (2003) also says that after 24 hours, successful self repair is
assumed.
MARIN bases their “self repair function” on statistics from the Dutch coastguard.
They established a function which has to be multiplied by the average number of
drifters per year (56.5 was obtained from statistics):
f(t)=1
f(t)=1/(1.5(t-0.25)+1)
for t<0.25
for t>0.25
where t = time after the engine failure occurred (hours)
The calculations are stopped after 24 hours. In the figure below the function is
illustrated for 56.5 drifters per year.
number of engine failures [1/year]
60
50
40
30
20
10
0
0
2
4
6
8
10
12
14
16
18
20
22
24
duration [hours]
Figure 3.5. “Self repair function” used by MARIN
with 56.5 drifters per year.
The probability that an emergency anchoring procedure is successful depends on
factors including the drift speed, the vessel size, the weather conditions and even
character of the sea bottom. Nevertheless, in order to harmonise their assumptions
GL and MARIN have changed the determining factor for emergency anchoring in
their models from the drift velocity to the wind speed. For mud/sand sea bottom
53
SSPA REPORT NO:
2005 4028
AUTHORS:
DNV, GL and MARIN will use proposal 2 from Figure 3.6. If the sea bottom is of
a different type the function has to be adapted and will be somewhere between
proposal 1 and 2.
1
probability of failure to anchor
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0
0.5
1
1.5
2
2.5
3
3.5
drift velocity [kn] (6% wind velocity)
Proposal 1 Probability of anchor failure
Proposal 2 Probability of anchor failure
Figure 3.6. Probability of failure to anchor (data from SAFESHIP 2005).
3.4.1.3
Further work
- Automatic Identification system (AIS)
GL has tried to include the reduction of the collision risk resulting from the use of
AIS in their Bayesian network model. This reduction results from the use of AIS
transponders installed on the wind farm to warn the crews of the passing ships of
the wind farm’s existence, and the AIS receivers to warn the personnel of the
wind farm of a ship on collision course with the wind farm so that the crew of that
particular ship could be warned. Under the assumption that 70% of ships are
equipped with AIS and use it, 68% of the potential powered collisions are likely
to be prevented. Under the assumption that 50% of the ships are equipped with
AIS, GL calculated a 50% reduction in collisions.
- Vessel Traffic Monitoring (VTM)
GL have also used included VTM in their Bayesian net model. From their study it
can be concluded that VTM can have a positive impact on the causation factor.
- Salvage tugs
Tugs can have a positive impact on the risk of collision for disabled vessels, but
there must be enough time available. It takes some time to issue an alarm and
activate the salvage tug. The tug then needs time to reach the vessel, and this is
dependent on the distance between the salvage tug and the disabled vessel, the
drift velocity of the disabled vessel and the cruising speed of the tug (both
dependent on the weather conditions). More time is then needed to establish a
54
SSPA REPORT NO:
2005 4028
AUTHORS:
towing connection between the vessels, a factor which depends on the equipment
available on the tug and the disabled vessel, the experience of the crews and the
weather conditions. When the towing line is established, the tug can reduce the
drift velocity of the drifting vessel. GL uses a conservative value of 1 hour for
stabilising the vessel.
The required towing capacity of the tug boat is calculated by similar formulas for
the forces as for the drifting speed. The coefficients are taken for a relative wind
direction of 155°, which is an angle based on experience and simulations. The
resulting coefficients are:
CdWind
=0.75,
C dWe
=0.5 and
C dcurr
=0.6
The required bollard pull is then calculated by the sum of the wind, wave and
current forces divided by 0.6 which is the average efficiency of the towing
capacity (the efficiency of the propeller can be reduced by factors such as heavy
seas). According to GL, the risk reduction factor attributable to tugs is in the range
of 3-14, which is probably a value only applicable to the coasts of Germany where
there is a high density of salvage tugs.
3.4.2
Ship – Wind Turbine Structure Consequence Estimation
Techniques
A vessel colliding with a wind turbine structure can result in damage to both the
vessel and the wind turbine structure, and consequences can include the
following:
1. Environmental damage:
• Spill of fuel oil from the vessel
• Spill of environmentally hazardous cargo from the vessel
• Spill of oil and hazardous liquids from the wind turbine structure
(gear-box oil, etc.)
2. Human injuries or fatalities:
• Human consequences are possible if the vessel founders or sinks or
if portions of the wind turbine structure fall and strike the vessel.
3. Economic loss:
• Loss of revenue resulting from temporary loss of power generation
capacity
• Potential loss of good will or reputation for companies involved
(especially if the accident is deemed the result of negligence).
• Damage to the ship which results in delay of the ship, costs due to
repair and stays at a repair yard, etc.
• Costs for salvage
55
SSPA REPORT NO:
2005 4028
AUTHORS:
Navigational risk assessments that have been carried out to date have focussed
primarily on consequences relating to spills, and the analyses have been rather
limited. More recent analyses (e.g. the navigational risk assessment carried out by
DNV for the Rødsand 2 Wind Farm (DNV 2007)) have included an assessment of
human consequences but these have been quite broad estimates.
The two types of scenarios to be considered when assessing the consequences of a
vessel impacting with a wind turbine structure include:
•
•
Vessel drifting into the wind turbine structure, either head-on or at an
angle
Vessel collision while vessel is powered, possibly at full speed, either
head-on or at an angle.
Factors that affect the severity of ship collision with a wind turbine structure
include:
•
•
•
•
•
Vessel weight
Vessel speed
Vessel stiffness
Wind turbine structure dimensions and materials
Wind turbine structure foundation type.
For conducting an analysis of collision severity, representative vessel types should
be selected for the analysis, based on vessel traffic statistics for the area.
Consideration should be given to those vessel types that may result in significant
environmental damage as a result of potential fuel oil spills or cargo spills. In
addition, “worst case” should be considered from the perspective of the wind
turbine structure. In Germany, the Federal Environmental Agency (UBA)
proposed a single hull 160,000 DWT oil tanker as the design ship in the accidental
limit state (ALS) to determine necessary preventative action in the event of an
offshore wind turbine failure (Biehl and Lehmann, 2006). For the Cape Wind
Farm assessment, vessel traffic information for the area and information on
navigation routes was consulted in conjunction with information on water depths
to select the vessels that could reasonably be expected to be involved in a collision
with the proposed wind turbine structures (Ali and Zheng, 2003).
For collision modelling, the angle of collision may have a significant impact on
the type of damage that the vessel may sustain. An angled impact may be more
severe than a head-on or a right angle impact and a sensitivity assessment of the
impact angle should be considered.
The main types of consequence modelling that have been carried out for offshore
wind farms include the following:
56
SSPA REPORT NO:
2005 4028
AUTHORS:
•
•
•
Semi-quantitative assessment: Scenarios are developed based on
probability modelling. Ship types that contribute the most to the collision
frequencies are considered in more detail in an oil spill analysis. Possible
spill volume is estimated based on bunker oil volume, number of bunker
oil tanks, and location of tanks.
Estimation of impact energy distributions for the structures: This is a
simplified method that calculates total kinetic energy for collisions using
displacement and velocity of vessels from local survey data.
Finite Element Modelling: The collision is simulated using finite element
analysis software to predict the collision sequence and specific damage to
both the ship and the wind turbine structure.
Each of these methods is discussed in more detail in the following sub-sections.
3.4.2.1
Semi-quantitative assessment
Semi-quantitative assessments usually involve a review of the probability
assessment and ship traffic data to identify those ship types that contribute most to
collision frequencies or that could be considered a worst-case in terms of amount
of oil spilled. The analysis described by Randrup-Thomsen et al. for Horns Rev is
an example of this technique. Oil spill scenarios for specific ship types were
developed based on the results of the ship collision frequency for the area.
Possible spill volume was estimated based on information taken from Lloyds
Register of Ships to develop a connection between the ship size and type and the
bunker volume in the tanks. A ship drifting sideways into one of the wind turbines
was considered to be the most likely collision scenario, and based on this it was
assumed that 30% to 50% of the total bunker oil volume would leak from the
tanks of each ship size. This was considered to be a conservative approach.
3.4.2.2
Estimation of Impact Energy Distribution
Collision software such as COLLRISK and COLLIDE generate impact energies
for vessel types. COLLRISK, for example, bases its analysis on the following
general equation:
E = ½ m (1 + a) v2
where,
E = total kinetic energy (kJ)
m = displacement of the vessel (tonnes)
a = hydrodynamic mass factor
v = velocity of the vessel (m/s)
(from Anatec, 2002).
57
SSPA REPORT NO:
2005 4028
AUTHORS:
This method was used in the Burbo Bank wind farm navigational risk assessment
(Anatec, 2002).
For the Gunfleet Sands Wind Farm, the Collide 2.60 model was used to create a
log-log plot of the annual collision frequency versus impact energy for locations
judged to have the highest and the lowest annual collision frequencies (Safetec,
2002).
These reports did not provide any further analysis with regards to possible ship
damage or spill volumes.
3.4.2.3
Finite Element Modelling
Finite element modelling develops a model of the collision and provides expected
damage to the ship structure and the wind turbine structure. Detailed input data is
required, including the following:
• Ship particulars, including materials information, sufficient to develop a
finite element model
• Wind turbine structure and material information to develop a finite
element model
• Soil condition information
Assumptions and simplifications are often made to be able to develop a model at a
reasonable cost and to suit the requirements of the analysis. Although FEMmodelling provides the most information on the collision sequence, it is expensive
due to the amount of time required, and the results are specific to the types of
ships modelled. However, this may be something that could be warranted in cases
where probabilities are relatively high or if there is a concern about a specific ship
type or cargo.
Examples of finite element modelling used to assess consequences of ship
collisions with offshore wind turbine structures are as follows:
Cape Wind Farm Analysis
An impact analysis model that used a “three degree of freedom dynamic impact
analysis computer program that solves Newton’s Second Law (i.e. Force equals
Mass times Acceleration) over time” was used for the Cape Wind navigational
risk assessment (ESS Group 2003). The largest ship size investigated in this
analysis was a 1500 DWT ferry, which was considered to be the largest vessel
that could possibly collide with the wind turbine structures (they were located on
a shoal). A computer program was developed using Matlab to carry out the
analysis. The program was referred to as “a multi-degree of freedom dynamical
impact analysis”. There was not very detailed information provided about the
foundation type – the tower was considered to have a spring at the soil interface
for the purposes of carrying out the impact analysis.
58
SSPA REPORT NO:
2005 4028
AUTHORS:
Biehl and Lehmann LS-DYNA Modelling of North Sea and Baltic Sea Wind
Farms
Biehl and Lehmann (2006) carried out a quantitative analysis of several collision
scenarios for different ship types and for 3 types of foundation structures (a
monopile, a jacket, and two tripod foundations (North Sea and Baltic Sea
locations). Soil properties varied significantly between the two locations. The
collisions were modelled numerically using finite-element modelling and
calculations were performed using LS-DYNA software. It was felt that due to
shortcomings in the modelling procedure the results could not be considered as
an exact representation of an actual accident. However, it was felt that they may
show possible consequences of such an occurrence and were useful for
developing preventative and response measures.
The numerical model developed by Biehl and Lehmann (2006) had two main
parts:
• Offshore wind turbine: this includes the structure, the foundation, and the
surrounding soil
• Ship and the surrounding water.
The two elements that are in direct contact during the collision, the wind turbine
and the ship, were represented as finite element models. The actual contact area
on both structures was modelled in more detail than other parts of the structure.
The foundation soil was considered to be an elasto-plastic deformable body.
Three foundation types were modelled, as follows:
•
•
•
Monopile: considered the most cost-effective foundation type, and is the
preferred solution in areas with sandy soil and water depths up to 25 m. It
does not offer much resistance in a ship collision scenario.
Tripods: These are used primarily in areas with water depths greater than
25 m. There are three piles, and Biehl and Lehmann claim that the local
stiffness of the diagonal is much higher compared to those of the jacket,
and this results in higher resistance to structural failure.
Jacket: May be used in water depths of 25 to 50 meters, and has higher
global stiffness as compared to the monopile. Apparently it exhibits a large
variation of failure modes during collision. The jacket structure is placed
on four piles.
59
SSPA REPORT NO:
2005 4028
AUTHORS:
Figure 3.7. Offshore wind turbine support structures considered by Biehl and
Lehmann (2006B). Figure taken from Biehl and Lehmann (2006B).
Four ship types were modelled, as follows:
• Medium size double hull tanker of 31,600 DWT
• Large single-hull tanker of 150,000 DWT
• 2300 TEU Container Ship
• Bulk carrier: 170,000 DWT
Results were as follows:
• Monopile: No serious ship damage or threat to the environment for the
double hull tanker and the 2500 TEU container ship. For the single-hull
tanker, however, a collision at an angle of 60 degrees caused the ship to
fail at the contact area and develop a large hole in the side structure. It was
estimated that this damage may allow the cargo from 2 holds to be
released.
• Jacket: For the double hull ship, the estimated damage is more severe than
for the monopile, but it is not considered to be enough to present a danger
of spill.
• Tripod: For the double hull tanker, it was estimated that the ship may have
severe damage resulting in penetration of both the outer hull and the inner
hull. This was expected to happen only if the ship comes into contact with
one diagonal during the collision sequence. If the ship does not hit a
diagonal strut, the consequences of the collision are expected to be similar
to those with the monopile. It was concluded that if the central joint is
placed low enough (deep enough in the water) to prevent contact with the
ship, collision consequences would be similar or better than what is
observed with the monopile.
60
SSPA REPORT NO:
2005 4028
AUTHORS:
Biehl and Lehmann concluded that the monopile and jacket foundation types
should be considered before the tripod, unless the central joint of the tripod could
be located lower than the maximum draught of ships expected to be trafficking the
area. The modelling was only carried out for a drifting collision. A powered
collision case was not modelled.
3.4.2.4
Environmental damage
The models developed by MARIN, DNV and GL include consequence models for
the outflow of oil and chemicals. For MARIN’s analysis of the Egmond aan Zee
wind farm (Kleissen, 2006), assumptions for ship damage and oil spills were
based on estimates of kinetic energy at the time of the collision. It was assumed
that all energy would be absorbed, and damage to the ship was calculated “based
partially on experience and partially based on complex calculations” (Kleissen,
2006). It was assumed that no spills would occur in the case of a drifting collision
(Kleissen, 2006).
DNV has developed crude oil outflow models for different accident types and
different hull configurations with normalised cumulative probability distributions.
GL also uses probabilistic and empiric formulas to calculate the oil spills.
The next step of a consequence analysis of a potential oil spill scenario may
include drift, spreading and dispersion modelling of the oil spill. Drift modelling
tools are regularly used in real response operations and may be combined with
GIS-based coastline environmental sensitivity data to estimate the potential
ecological damage and the beach clean up resources required and associated costs.
The spill conditions, beach contamination and environmental impact vary from
case to case and it is difficult to predict and to generalise in monetary terms the
magnitude of potential environmental consequences.
The SeaTrack Web, developed by SMHI (Swedish Meteorological and
Hydrological Institute), is a well established GIS-based modelling tool for drift
predictions of oil and chemical spills at sea.
3.4.2.5
Human consequences
MARIN’s risk analysis for Egmond aan Zee wind park only estimated potential
injuries or fatalities resulting from the wind turbine structure falling on the ship’s
deck, and did not estimate injuries from other collision aspects (Kleissen, 2006).
Estimates of the probability of a wind turbine collapse were based on kinetic
energy estimates for specific ship types and operating speeds, as was done for the
environmental damage estimates. Worst-case scenarios were assumed so that
estimates would be conservative. For example it was assumed that in the case of a
61
SSPA REPORT NO:
2005 4028
AUTHORS:
collapse, the wind turbine structure would always fall on the ship rather than away
from it, and that the pile and the rotor would land completely on the deck, rather
than grazing it. No dynamic modelling or experiments were carried out to validate
assumptions.
DNV calculates the loss of lives based on statistics but states that the statistical
uncertainty becomes large, because no accidents of this type have occurred. The
basis for the detailed calculations was not provided, but the ship-ship collisions
gave the biggest contributions. Considering the “worst-case” approach for the
calculation of fatalities it was concluded for all calculations that the risk of dying
due to collision with the wind farm is not significant.
3.4.2.6
Economic loss
DNV states that the loss of property is related to salvage, transfer or loss of cargo,
repair cost and loss of profit due to downtime. In DNV’s report for Rødsand
(Christensen 2007), the expected costs are taken from information given by
insurance companies and classified for different ship types. These values are then
related to the most common ship sizes and connected to the probabilities for the
different losses.
The German authorities (BSH) will include an assessment for the consequences of
a collision in their newest guidelines. This assessment has to describe the safety of
the foundation and the head bearing of the wind power plant (Biehl, 2007). Biehl
also commented on the ongoing research project “Collisions of Ships and
Offshore Wind Turbines: Risk of nacelle impact”. Too many assumptions have to
be made to include nacelle impact in the guidelines, because probabilities and
consequences are unknown for many factors (e.g. forces acting on the head
bearing, probability of a fall off, falling speed and direction of the hub or turbine,
braking effect of the decks and cargo of the ramming vessel, etc.).
3.4.3
Ship to Ship Collision Probability Estimate
The wind park may affect the shipping traffic and can have positive or negative
impact on the risk of ship-to-ship collision. The ship traffic routes might be
modified due to the location of a wind park. Traffic separation schemes, improved
buoyage and similar changes could lead to a reduced probability of ship-ship
collisions, while changing the positions of existing shipping lanes, compression of
the traffic, additional crossing of ships over main shipping lanes or other
modifications could lead to increased probabilities of such a collision.
The probability is in general calculated for all shipping lanes by:
F(ship-ship collision)=P(collision| encounter)*F(encounter)
Many models exist to calculate the risk of ship-ship collisions. An investigation
into the details of these models goes beyond the scope of this project. For further
62
SSPA REPORT NO:
2005 4028
AUTHORS:
information on ship-ship collision models other references should be consulted
(for example see Kristiansen (2005)).
3.4.4
Ship to Ship Collision Consequence Estimate
Consequences of ship to ship collisions which could be caused by navigation
changes resulting from offshore wind parks include injuries, loss of life,
environmental damage, cargo loss, and ship damage.
In terms of consequences the ship-to-ship collision might be worse than a shipturbine collision, because usually higher kinetic energies are involved which lead
to more extensive damage and the crew, cargo and bunker of two ships instead of
one ship might be affected and might cause more fatalities, economic and
environmental damage than the ship-turbine collision.
The consequences of ship-to-ship collisions resulting from the presence of a wind
farm would be the same as for collisions caused by other factors such as
equipment malfunction, navigational error, human error, etc. Historical accident
databases such as LMIU can be consulted to provide empirical data on
consequences from actual collisions. This type of detailed investigation of
consequences has not been included in most navigational risk assessments for
offshore wind farms, and changes in probability of ship to ship collisions are
generally the extent of the work that is usually performed for this area.
If a ship-to-ship collision results in a spill that drifts towards an offshore wind
farm, consideration should be given to the additional challenges of spill clean-up
and recovery in a wind farm area. The Bonn Agreement Counter-Pollution
Manual (Bonn Agreement, 2007) states that wind turbines should be turned off
when recovery vessels are operating within a wind park, and that dispersant
spraying, if appropriate, would need to be done from a vessel and not an aircraft.
3.4.5
Grounding Probability Estimate
Offshore wind farms may actually have a positive effect on the risk of grounding,
as they may result in a reduced probability of grounding in the area of the farm.
At the same time, the probability of groundings might be increased due to
modifications of the shipping lanes. The potential of the grounding risk to
increase or decrease significantly should be looked at on a case by case basis.
Several models are used for risk analysis concerning grounding of ships and the
discussion of these risks goes beyond the scope of this work. The technical
literature is again referred to if more detailed information is desired.
63
SSPA REPORT NO:
2005 4028
AUTHORS:
3.4.6
Grounding Consequence Estimate
If it is the case that groundings are increased, consequences may need to be
considered. Historical accident databases such as LMIU can be used to obtain
empirical data on consequences from actual groundings. It is unlikely that detailed
investigation of consequences from groundings would be necessary for
navigational risk assessments for offshore wind farms. For a zero-alternative risk
analysis, the grounding consequences are of importance.
3.5
Effects on Radar, Radio, Navigation Equipment,
etc.
A number of studies have investigated the effects of wind turbines on
navigation equipment, as follows:
•
North Hoyle Wind Farm: Experimental field tests were carried out to
assess the effects of the wind farm structures on marine systems in
operational scenarios. The work was commissioned by the UK
Maritime and Coast Guard Agency, Navigation Safety Branch, and was
carried out by QinetiQ (Howard and Brown, 2004). A summary of the
goal of the work and results was provided by Howard and Brown
(2004) as follows:
“The trials assessed all practical communications systems used at
sea and with links to shore stations, shipborne and shore-based
radar, position fixing systems, and the Automatic Identification
System (AIS). The tests also included basic navigational equipment
such as magnetic compasses. The effects on the majority of systems
tested by the MCA were found not to be significant enough to affect
navigational efficiency or safety, and an on-going collection of data
on such systems is expected prove these conclusions.”
Additional work was carried out at North Hoyle in March 2005 to
investigate the effects on aircraft systems (Brown, 2005). Tests were
carried out with a Sea King Mark II aircraft, and results indicated
that “radio communications from and to the aircraft operated
satisfactorily, as also did its VHF homing system” (Brown 2005).
•
Horns Rev (as reported in ESS Group, 2003): It has been reported that
there have been no disruptions or difficulties observed with VHS
communications between vessels in and around the wind park, or
between vessels in and around the wind park and the traffic
coordination centre at Esbjerg and the Coast Guard/Rescue Centre. In
addition there have been no radar shadows observed from the towers’
64
SSPA REPORT NO:
2005 4028
AUTHORS:
rotating turbines in the Horns Rev park.
•
Egmond aan Zee Wind Farm (OWEZ): As part of the risk assessment
for the OWEZ monitoring and evaluation program (Kleissen, 2006),
MARIN carried out simulations to assess the effects of the planned
wind farm on shipping radar. The simulations were performed on
MARIN’s full mission bridge simulator, and three runs were
performed. Radar observations on a containership were simulated, and
a coaster and a tug were used as radar targets. The study concluded that
the wind farm has a negative influence on radar performance, but not to
the level that detection of other vessels becomes impossible. It was also
concluded that radar performance improves when the number of wind
turbines between two ships decreases, as would occur when ships are
sailing to the same corner. The study did not consider the impact of
ghost targets resulting from reflections of side bundles, because the
simulator was not equipped to include the reflections. It was
recommended that a field trial be conducted after completion of
construction of the wind farm to assess this issue.
3.6
Effects on Search and Rescue Operations
Possible speculated effects of an offshore wind park on search and rescues
(SAR) operations include both positive and negative influences. Positive
influences include the establishment of a place of refuge at each wind turbine
structure. A negative influence is that any collision with wind turbine structures
will add additional cases to the coast guard’s SAR work load. Another potential
negative influence is interference of wind farm structures with search and rescue
helicopters.
An assessment was carried out for Cape Wind Farm (ESS Group, 2003) in
Massachusetts, USA, by reviewing information from the coastguard’s database
of missions, by reviewing USCG SAR operational guidelines, and through
consultation with coast guard staff involved in SAR. Data from a ten-year period
for the area around the proposed wind farm was evaluated. The majority of the
responses were by sea, although 4% were by air. The study concluded that the
presence of the wind farm would be a benefit for search and rescue for the
following reasons:
•
each wind turbine structure would have an alphanumeric identifier
painted on it, and the coast guard and other rescue agencies would have
a plan showing the location of each tower, thus helping them with
planning rescue operations.
65
SSPA REPORT NO:
2005 4028
AUTHORS:
•
the wind turbine structures would have rescue lines attached and could
serve as a place of refuge for persons in the water after abandoning ship
or falling overboard.
•
work vessels would be in the area periodically for wind farm
maintenance and these vessels would be able to assist vessels in distress
in the area.
The UK’s Maritime and Coast Guard Agency undertook helicopter trials in
March 2005 at the North Hoyle offshore wind farm to investigate whether
marine and shore-based radar systems would be adversely affected by the
presence of an offshore wind farm (Brown, 2005). The study also included a
discussion of how search and rescue helicopters may be affected by the
presence of offshore wind farms. Results showed that radio communications to
and from the aircraft were satisfactory; vessels, turbines, and personnel in the
wind farm could be clearly identified in dry weather on the aircraft’s thermal
imaging system; and there were no compass deviations (Brown, 2007). Some of
the issues identified were as follows:
there are “significant radar side lobe returns from structures” (Brown,
2005), and these can limit detection of vessels that are within 100 metres
of the turbines;
• turbine blades must be confirmed to be locked before helicopter rescue
can be considered safe. The North Hoyle wind turbine blades could not
be remotely locked and thus helicopter rescue from these turbine
structures was considered extremely (and perhaps prohibitively)
dangerous;
• thermal imaging is limited when there is mist or precipitation;
• vessel or shore-based marine radar tracking of helicopter movements
within the wind farm was poor;
• aircraft power requirements are increased downwind of the wind farm.
There were also limitations identified for the specific SAR helicopter,
equipment, and crew in the study area. The Sea King Mark III helicopter used
has a radar console that was not visible from the cockpit, and the radar
operator doubled as a rescue hoist operator. This meant that surface rescue
from the helicopter would not be feasible in restricted visibility conditions, as
the crew would be “radar blind” when operating the winch.
•
3.7
Validation of Risk Assessment/Quality
Control/Uncertainty/Sensitivity
Available data for validation could include empirical accident and incident data
from contacts with offshore structures, and comparison with risk assessments
66
SSPA REPORT NO:
2005 4028
AUTHORS:
carried out for other wind farms. A discussion of model uncertainty and
validation is provided below.
3.7.1
Model accuracy and uncertainty
A number of uncertainties are introduced when risk calculation models are
elaborated. Various degrees of uncertainty are associated with the following areas
and factors.
•
•
•
•
•
Ship traffic statistics – recorded AIS ship traffic data have high accuracy
but need to be simplified for modelling purposes to a limited number of
main shipping lanes and all ships do not follow the lanes
The outlined risk model
–
there may be potential accident scenarios that are not included or
unknown secondary hazards
limitations in the possibility of describing the reality in a model
–
Engineering judgements and assumptions on key model probability
parameters
Assumptions regarding the consequences of collision accidents in terms of
fatalities per final outcome, potential environmental impact and economic
damage value.
Statistical and empirical probability data - historical data on collision
probabilities are incomplete and reflect historical safety regimes and
technical standard of ships
If the range of uncertainty for each parameter is estimated, the possible impact of
the uncertainties and needs for further information and analysis may be identified.
Sensitivity analysis can be conducted by systematically varying some key
parameters in the calculation of the final outcome. In the case study for Kriegers
Flak (Chapter 5) this is exemplified.
3.8
Risk Acceptability and Risk Acceptance Criteria
There are no internationally adopted general standards for risk acceptability
applicable to the issue of navigational safety and offshore wind farms. There are
also no quantitative acceptance criteria established in Sweden that may be
applied in this area.
The formulation of quantitative accident risk acceptance criteria is a sensitive
political issue and very much associated with the subjective perception of risk
and risk aversion.
67
SSPA REPORT NO:
2005 4028
AUTHORS:
The qualitative risk estimations addressed in this study basically focus on the
probability of ship wind turbine collision and are expressed in probability per
year or expected return period (in years) for the event. The consequences of the
collision events are not quantitatively modelled in detail and no evaluation in
relation to possible or proposed acceptance criteria are presented.
3.8.1
Individual and societal risk
When both the probability and potential consequences of accidental events are
analysed, the combined risk figures are usually quantified in terms of expected
fatalities. Acceptance criteria can then be formulated either based on individual
risk or societal risk.
For specific occupations, locations or activities, individual acceptance criteria may
be expressed by an annual fatality risk. For large systems, which expose a large
number of people to risks, and where a large number of people are affected by
possible accidents, societal risk considerations provide a more appropriate basis
for risk acceptance criteria. The societal risk is expressed in terms of frequency
versus number of fatalities, and two of the most commonly used methods of
describing such risks are risk matrices or FN-curves. Risk matrices and FN
diagrams will also indicate which levels of risks are acceptable and which are not.
Potential Loss of Lives (PLL) is another measure of societal risk for a defined
system or activity.
Table 3.10. Example of Risk Matrix.
Example of Risk Matrix with Risk index figures and indicative acceptance
criteria. Red area is unacceptable risks, Green is acceptable
Severity index
Frequency index
Minor Significant Severe Catastrophic
1
2
3
4
Frequent
7
8
9
10
11
Probable
6
7
8
9
10
Reasonable probable
5
6
7
8
9
Remote
3
4
5
6
7
Extremely remote
1
2
3
4
5
The area between the red intolerable area and the green tolerable area is called the
ALARP area (As Low as Reasonably Practical) and indicates that risk reduction
measures should be applied. The scale of consequences illustrated by the risk
matrix above may also be transformed into terms that represent environmental
consequences (e.g. volume of spilt oil) or economic loss figures.
68
SSPA REPORT NO:
2005 4028
AUTHORS:
3.8.2
Individual risk acceptance in shipping industry
With regard to maritime safety and acceptable risk exposure of crew members,
risk acceptance criteria have been proposed. The criteria proposed in MSC 72/16,
based on figures published by the UK Health & Safety Executive, have been used
by various FSA studies. The table below presents the suggested acceptance levels
for the individual risk to crew members.
Table 3.11. Individual risk levels for exposed crew members.
Individual risk levels for exposed crew members
Risk level
Annual fatal risk
Maximum tolerable risk for crew members
10-3
Negligible risk
10-6
3.8.3
Different types of criteria for offshore wind farms
As illustrated in Chapter 3.2, a number of different views on risk acceptance for
offshore wind farms can be identified, and relevant acceptance criteria may
consequently be formulated in different terms.
In this study, the navigational safety perspective is the main focus and criteria
may be based on relative risk evaluation. For example a criterion could be that the
park establishment shall not generate increased probability for ship-structure
collision, ship grounding, ship stranding or ship-ship collision in a specific
navigational area where the farm is located.
If the criterion is formulated in absolute risk figures, e.g. by ALARP limits, then it
is still important to compare the risk figures with the baseline case before the wind
farm is established.
In many cases it is also relevant to study the risks from the proposed wind farm’s
point of view. This is of course relevant for the farm owner and operator and may
for example influence insurance discussions. Calculation of expected collision
probability or return periods is also relevant for comparing different wind farm
layout or localisation alternatives from a navigational safety point of view.
3.8.4
Acceptance criteria in Germany
The German authorities have agreed with a group of experts on risk acceptance
criteria. Offshore wind farms that result in a collision probability with a return
period of more than 100-150 years are generally accepted by the German
69
SSPA REPORT NO:
2005 4028
AUTHORS:
authorities. This range of return periods covers the results from different models
(i.e. no specific model type is specified). A precaution and safety concept might
still have to be developed for the facility. If the collision frequency of crude oil
tankers and chemical or product tankers plays a major role in this collision
frequency, further studies might be necessary. If the return period is between 100
and 50 years the project might be rejected. Further studies are required and a more
detailed look into possible consequences is mandatory. Return periods below 50
years are generally unacceptable and the wind farm may only be considered for
possible approval from the authorities if risk reducing measures which increase
the return period above 50 years or more are applied (Bundesministerium für
Verkehr-, Bau und Wohnungswesen, 2005).
Table 3.12. Risk acceptance criteria in Germany (based on data from
Bundesministerium für Verkehr-, Bau und Wohnungswesen (2005)).
Acceptance
Time between collisions
[years]
Acceptable
>100 (100-150)
Further analysis necessary:
acceptability considered on a
case by case basis,
50…100
Not acceptable
<50
With regards to oil spills or other pollution resulting from a collision of a ship
with a wind power plant the park can receive approval, if the return period is
between 300 to 450 years for a spill volume of 50 m3 or more.
The group of experts in Germany (Bundesministerium für Verkehr-, Bau und
Wohnungswesen, 2005) gave also orientation values for different assumptions
made for the calculations. The German risk acceptance criteria presented above
are valid only in combination with these orientation values. This includes a
minimum distance from shipping lanes to a wind park of 2 nautical miles plus the
500 meter safety zone (safety zone according to UN law of the sea convention
article 60). Other values specified by the experts are as follows (for more values,
see Bundesministerium für Verkehr-, Bau und Wohnungswesen, (2005)):
•
The minimum ship size to be looked at: 500 GRT
•
the maximum drift speed of disabled ships: 4 kn
•
average ship speed: 11 to 18 kn for RoRo; about 20 kn for RoPax; 25 kn
for large container vessels; and 35 kn for High Speed Craft (HSC)
70
SSPA REPORT NO:
2005 4028
AUTHORS:
•
lateral distribution should consist of a Gaussian distribution and a uniform
distribution with a size of 2% of the Gaussian distribution (the width of the
uniform distribution is assumed to be 6 times the standard deviation)
•
Orientation values for different fairways are:
Table 3.13. Standard deviation for the Gaussian distribution describing the lateral
distribution of ships on the lanes (Bundesministerium für Verkehr-, Bau und
Wohnungswesen, 2005).
Description
Standard deviation for
Gaussian distribution [nm]
Port approach
0.2 to 0.3
Conspicuous navigational points,
e.g. navigational marks, buoys
0.3 to 0.4
separation
0.5
0.5 to 1.0
2.0
•
The relevant area to be looked at for powered collisions is 15 nm
(20nm) from the outermost wind power plants.
•
The causation factor (probability for a ship on a collision course to not
take any corrective action, due to technical or human failure) is
3.0E-04.
•
The effective collision breadth is 1.2B plus the diameter of the obstacle
for powered collisions and the ship length (90 degree drift direction)
plus the obstacle width for drifting collisions.
•
the maximum drift time is 24h.
•
The rudder/ engine failure rate is 2.5E-04 per hour. For ships with
double engine-rudder installations this factor can be reduced.
•
The probability of anchor failure is dependent on the wind speed, waves
and sea bottom characteristics. The following probabilities are assumed
for Baltic Sea conditions:
71
SSPA REPORT NO:
2005 4028
AUTHORS:
Table 3.14. Probability of anchor failure (Bundesministerium
für Verkehr-, Bau und Wohnungswesen, 2005).
•
Beaufort scale
Probability of anchor failure
0
0.01
1
0.01
2
0.01
3
0.01
4
0.035
5
0.07
6
0.126
7
0.21
8
0.35
9
0.49
10
0.63
11
0.7
12
0.7
The following formula can be used for the failure rate depending on the
time required to repair the failure:
f(t)=1
for t<0.25 h
f(t)=1/(1.5(t-0.25)+1) for t>0.25 h
3.8.5
•
When including effects of AIS an efficiency is assumed with a factor of
1.25
•
When including effects of VTM an efficiency is assumed with a factor
of 2-4
Other acceptance criteria
In GL’s risk analysis report for the Belgian wind farm “Thornton Bank” the risk
matrix and values shown in the table below are used for defining the qualitative
frequency, which is the background for acceptance/ non acceptance.
72
SSPA REPORT NO:
2005 4028
AUTHORS:
Table 3.15. Classification of consequence severity and occurrence frequency (Neuhaus and
Thrun 2003).
Frequency H [1/year]
safety (quantitative)
H>10-1
10-1≥H>10-2
Frequency H [1/year]
environment
(quantitative)
H>2x10-1
2x10-1≥H>2x10-2
probable
improbable
frequent
minor
Frequency
(qualitative)
Consequence / failure
severity classification
10-2≥H>10-3
H≤10-3
2x10-2≥H>2x10-3
H≤2x10-3
remote
extremely remote
extremely
improbable
major
severe
catastrophic
In MARIN’s study by Kleissen (2006) a so-called “orientation” value is used for
societal risk; the frequency of 10 people dying per seaway (per kilometre) is
allowed to be a maximum of 10-4. The “orientation” value is taken from the riskstandards for the transport of dangerous goods and it is stated by Kleissen (2006)
that it is questionable whether or not this standard can/may be used for their study.
DNV has proposed risk acceptance criteria for application in the marine industry.
The criteria are neither official DNV criteria nor are they recognized by regulatory
bodies. The criteria are as follows:
Table 3.16. Proposed individual human fatality risk acceptance criteria for the shipping
industry (Spouge 1997 and DNV 1999).
Risk acceptance criteria
Value
Maximum tolerable risk for crew members
1 fatality per thousand at risk per year
Maximum tolerable risk for passengers
1 fatality per ten thousand at risk per year
Maximum tolerable risk for public ashore
1 fatality per ten thousand at risk per year
Table 3.17. Proposed total loss, cargo spill and bunker spill targets for the shipping
industry (DNV 1999).
Risk Targets
Value
Target total ship loss frequency
2 losses per thousand ship years
Target cargo spill risk
20 tonnes per million tonnes transported
Target bunker oil spill risk
20 tonnes per million tonnes consumed
73
SSPA REPORT NO:
2005 4028
AUTHORS:
Swedish Rescue Services Agency presents principles or general starting points for
design of risk criteria. These four principles are (Davidsson et al 2003):
•
•
•
•
Rimlighetsprincipen (principle of reasonableness)
Proportionalitetsprincipen (principle of proportionality)
Fördelningsprincipen (principle of apportionment)
Principen om undvikande av katastrofer (principle of avoiding
catastrophes)
3.9
Statements and recommendations from other
stakeholders
In this context it may be interesting to note that the German Nautical Association
proposes a safety zone of 1 000 meters around offshore wind farms instead of the
500 meters established by international law of the sea and agreed zones of
national jurisdiction at sea. The proposal is based on navigational safety
considerations (Deutscher Nautischer Verein 2004).
74
SSPA REPORT NO:
2005 4028
AUTHORS:
Risk Reduction measures
4
RISK REDUCTION MEASURES
This chapter includes a discussion of risk reduction measures that are appropriate
for addressing specific hazards and risks. In addition methods for evaluation of
risk reduction measures will be recommended.
Risk reduction measures can be grouped into the following main categories:
•
Measures to reduce the probability of accidents and incidents
•
Mitigation measures to reduce consequences (ship-related consequences,
environmental consequences, etc.)
Figure 4.1. Risk reduction measures.
Examples of risk reduction measures include improved tug boat response time,
installation of lights and navigation aids on wind turbine masts, and establishing
safety management plans. It may be possible to assess some measures such as
improved tug boat response time in a quantitative manner, while others, such as
those targeted at reducing human error, may be assessed qualitatively. Measures
can be grouped into categories based on the primary area of application, as
follows:
• measures that can be applied at the wind farm;
• measures for ships, including specific measures for those transporting
dangerous goods;
• measures to be applied to the whole marine area surrounding the park.
75
SSPA REPORT NO:
2005 4028
AUTHORS:
Examples of risk reduction measures in each category are provided in the
following list.
Risk reduction measures that can be applied at the wind farm level to reduce the
probability of a collision are as follows:
•
Optimising the assembly of the wind power plants. This can be done using
a risk analysis, even though the arrangement has a low impact on the
collision frequency.
•
Marking the wind farm as a prohibited area in the sea charts, the
aeronautical charts and the nautical handbooks.
•
Equipping the wind farm radar equipment and radar antennas with at best
with redundant transmission. Further possibilities are VHF, radio
frequency units, etc.
•
Installing equipment with navigation lights on each WPP.
•
Declaring safety zones around every WPP.
•
Installing AIS transponders (at least two) at the wind farm. Studies carried
out as part of the SAFESHIP project (SAFESHIP 2006) have shown that
the use of AIS equipment on wind farms and ships will result in a reduced
collision frequency.
•
Producing a safety manual and preparing emergency plans.
•
Installing camera and video equipment for observation of the wind farm
area.
Measures for the wind farm to reduce the consequences:
•
Constructing the WPPs in such a way that as little damage as possible is
inflicted on the ship during a collision. This includes ensuring that the
structural damage are kept low, and that the tower, the housing and the
rotor blades of the WPPs fall away from the ship in the event of a
collision.
•
Structural construction of the surrounding of the power plants in a
“collision-friendly” way, i.e. using fenders, etc.
•
Using environmentally friendly coolants for the transformers.
•
Equipping the WPPs with a fast shutdown and an emergency brake.
76
SSPA REPORT NO:
2005 4028
AUTHORS:
•
Installing the turbines at such a height that crews of colliding ships cannot
be hit by rotor blades.
•
Equipping the substation with a helicopter platform and installing docking
possibilities for salvage tugs and SAR boats.
•
Marking every power plant with a unique ID to simplify search and rescue
operations.
•
Producing a safety manual and preparing emergency plans.
•
The cables should run covered in the ground to minimise dangerous
situations in case of emergency anchoring
Measures for ships to reduce the probability of collision with wind farms include:
•
Equipping the vessels with AIS, redundant navigational equipment,
redundant propulsion systems, good conditions to make the connection by
ropes to a tug easy, and reinforced towing bollard.
•
Equipping ships with ECDIS (Electronic Chart Display and Information
System), to potentially reduce navigational errors
•
Education and training for the crew and preparedness for critical
situations.
•
Vessels which sail in the surrounding of the wind farm should be prepared
to execute an emergency anchoring.
Measures for the sea area to reduce the probability of a collision:
•
Establishing a traffic separation scheme.
•
Constant monitoring and observation of the area, of the passing ship traffic
and of the wind farm by Vessel Traffic Management. A study based on
empirical data [Safeship, 2006] showed that VTM might lead to a
reduction by a factor 2-10 of powered collisions. The German
harmonisation group states that a factor of 2-4 is realistic.
•
Observation and control of the vessels and their operation by the
authorities.
•
Emergency management
•
Monitoring and reporting mechanism for passing and drifting vessels. In
the Netherlands for example every disabled ship in the EEZ is obliged to
77
SSPA REPORT NO:
2005 4028
AUTHORS:
report to the Dutch coast guard immediately.
•
Allocation of salvage tugs including a certain sea position in heavy
weather for the tugs close to areas with a high frequency of collision and
grounding.
•
Clear marking of the modified shipping lanes to avoid ship-ship collisions
and groundings.
•
A boarding team should be accessible at all time to assist the ship crew in
establishing a connection to the salvage tug.
•
Updated sea charts showing the wind farm should be available as early as
possible to the public.
•
A safety zone should be announced around the wind farm.
•
It should be forbidden to sail through the wind farm.
Measures for the sea area to reduce the consequences:
•
The private or state-owned tug assistance needs to have the ability to reach
the disabled vessel in time and should have sufficient bollard pull to stop
the drifting of the disabled vessels.
•
Oil spill response plans should be made and sufficient capacities for
response should be available.
•
Regular training of SAR and oil spill response units should be conducted
in or close to the wind farm.
A navigation channel from the base harbour of the working vessel to the working
area together with a traffic controller coordinating the working vessels shall be
introduced.
For the most exposed turbines the gravity foundations should be designed so the
foundation plate is in level with the seabed. Alternatively scour protection or
similar should be made with a minor gradient towards the centre column in order
to avoid bottom rupture. None of the hazards related to the smaller boats were
found to be in the unacceptable region.
Because the probability for a collision between a wind power plant and the service
vessel is high, SAFESHIP suggests SWATH (Small Waterplane-Area Twin
Hulls) for the work.
78
SSPA REPORT NO:
2005 4028
AUTHORS:
From the risk analysis point of view, damage of the wind power plant is the most
probable consequence. The gearbox oil and the light diesel oil used in the power
plants should therefore be environmental friendly.
During the construction of the park special measures should be conducted:
•
Special sailing routes for the transport vessels.
•
Clear marking of the area where the power plants are to be constructed, i.e.
with lights, radar equipment, etc.
•
Providing notification designating the wind farm as an air obstacle before
the start of the construction.
•
Special safety and rescue plans during construction of the wind farm.
•
Safety notices to be sent out to ships which regularly sail within a
designated proximity to the wind farm construction site.
•
Informing the public and all the involved stakeholders in the area
4.1
Bonn Agreement Counter-Pollution
Recommendations
The Bonn Agreement refers to a mechanism by which North Sea states, together
with the European Community, work together to combat oil pollution in the North
Sea and to perform surveillance to detect pollution. North Sea states include
Belgium, Denmark, France, Germany, the Netherlands, Norway, Sweden, and UK
and Northern Ireland. The Bonn Agreement Counter-Pollution Manual
(www.bonnagreement.org) includes a chapter on offshore wind farms, which
recommends that response authorities be prepared to handle incidents where oil
slicks drift in to offshore wind farms. Chapter 8 of the manual discusses various
measures that can be taken to reduce risk of oil spills in the vicinity of wind
farms. Possible response mechanisms to a spill drifting towards a wind farm are
described as follows:
•
where mechanical recovery is feasible, recovery vessels will need to be
allowed in to the park. It is proposed that turbines should then be switched
off, even if there is adequate clearance between the ship and the rotorblade.
79
SSPA REPORT NO:
2005 4028
AUTHORS:
•
where the use of dispersants is recommended, the dispersant spraying
should be done from a vessel, and spraying aircraft would not be permitted
to operate within a wind farm.
Several preventive measures are also recommended in the Bonn Agreement
Counter-Pollution Manual, as follows:
“the definition of a safety zone around the area, use of mist horns, signalisation of
all structures at all times for nautical and aerial purposes, installation of oil
retention tanks, list-keeping of all ships operating on behalf of the owner of the
windfarm, numbering of structures, early warning of the authorities for all parkrelated activities in the shipping routes, the organisation by the owner of (multi-)
annual simulation exercises on various subjects such as nautical emergencies,
towing or pollution response, and the obligation on the owner (to be determined
case-by-case) to follow the requirements of the competent authorities with regard
to navigational requirements and safety.”
4.2
Evaluation of Risk Reduction Measures
If a quantitative risk analysis has been carried out for a wind farm, it should be
possible to quantitatively assess at least some of the risk reduction measures,
depending on the type of model that has been used for the risk analysis. For
example some estimates of the probability of drifting collisions include inputs on
tug response times – if these times can be shortened due to additional tugs or
faster tugs, then the effect on the probability of collision can be estimated.
Construction of an event tree using an initiating event such as “engine failure” or
“black out”, and developing branches to describe different chains of events
would be useful in identifying and evaluating risk reduction measures for a
drifting failure.
80
SSPA REPORT NO:
2005 4028
AUTHORS:
Case Study: Application of Selected Methods to Kriegers
Flak Proposed Project
5
CASE STUDY: APPLICATION OF SELECTED
METHODS TO KRIEGERS FLAK PROPOSED
PROJECT
The Kriegers Flak project was chosen as the site for a case study for this report.
This was because there have been two independently performed and documented
risk analyses carried out for the site using models which have been used for many
other risk assessments in Europe. The previous analyses were carried out by
Maritime Research Institute Netherlands (MARIN) and Germanischer Lloyd,
Germany (GL).
The reports available from GL (Otto and Petersen 2003, Povel et al. 2004 and
Otto 2004) and MARIN (van der Tak and Rudolph 2003 and van der Tak 2005b)
consist of one original report from GL and one from MARIN and reports with
updates, extensions, harmonisations and corrections of the original reports. All the
studies are for the offshore wind farm proposed in the German exclusive economy
zone (EEZ) of the Kriegers Flak in the Baltic (also called Kriegers Flak I).
The table below shows selected results from MARIN and GL concerning route
bounded ship traffic. The effect of emergency salvage is not included.
Table 5.1. Results from MARIN’s first report (van der Tak und Rudolph 2003),
their latest report (van der Tak 2005b) and from GL’s latest report (Otto 2004)
concerning route bounded ship traffic. The effect of emergency salvage is not
included.
Company
MARIN
GL
First report First report Latest report Latest report
Year 2000
Year 2010
Year 2010
Year 2000
Return period,
36
29
67
578
drifting [years]
Return period,
1344
1043
330
218
powered [years]
Return period,
35
28
56
158
total [years]
MARIN’s latest report only includes calculations for the year 2010, which is not
comparable with GL’s latest report which is for the year 2000. In MARIN’s first
report, the results for 2010 are about 80% of the results of 2000. Assuming that
the same relation could be applied to their latest report, the return periods for the
year 2000 should be 84 years (drifting), 413 years (powered) and 70 years (total).
Comparing these figures to GL’s, the following could be stated:
81
SSPA REPORT NO:
2005 4028
AUTHORS:
•
•
•
MARIN’s calculation is more conservative regarding drifting ships (GL’s
return period is about 7 times larger than MARIN’s).
GL’s calculation is more conservative regarding powered ships (MARIN’s
return period is about 2 times larger than GL’s).
MARIN’s calculation is more conservative regarding the total return
period (GL’s return period is about 2 times larger than MARIN’s).
The approach and structure of the risk analyses presented by GL and MARIN are
generally traceable. There are problems, however, with comparing the results
because of the different assumptions made for the location of the wind park, the
different traffic volume, the different shipping lane coordinates and the different
models which are used and which are not fully transparent in the reports. The
documentation of the calculations presented in the reports is often incomplete and
not fully transparent with regard to assumed numerical parameters, etc. On the
webpage of other companies involved in the project, the maps even show different
locations for the wind farm.
The objective of carrying out this case study was to try to find out why the results
calculated by GL and MARIN differ. The method used was to simulate or emulate
the two models with SSPA’s model as a starting point. When available, model
parameters and input data were compared. When data were available, the three
models mentioned above were also compared to Det Norske Veritas’ (DNV)
model “MARCS”.
It should be mentioned here that most models have undergone strong development
in the meantime, not least because of the new possibilities offered by statistical
processing of recorded AIS data, which is a corner stone for every state-of-the-art
navigational risk analysis.
Within the Baltic Master project, another case study for Kriegers Flak is
performed (see Baltic Master 2007).
5.1
Models
As previously described in the study, two different situations may result in
collision of ships with offshore wind farms. The first one is the powered collision,
in which a navigational error due to human or technical error in the navigational
instruments or a combination of the two leads to the ship sailing into the offshore
wind park if the error is not detected in time. The second one is the drifting
collision in which parts of the propulsion system (which includes the engines) are
affected by failure, the crew loses control over the vessel and the vessel starts to
drift. If the vessel drifts towards the park, it may drift into it depending on other
factors such as the wind speed and wind direction.
82
SSPA REPORT NO:
2005 4028
AUTHORS:
An overview of different models is given in the table below. The models of DNV
and SSPA Sweden AB are included in the comparison. Model parameters were
investigated systematically.
Table 5.2. Overview of models used for calculating collision frequencies.
Model
Hazard Identification
Collection of input
data
Cruising speed 6)
Powered model
Standard deviation
for course offset
Uniform distribution
in addition to
Gaussian distribution
for course offset
Course deviation
Causation factor/
Navigational Error
Rate
Dependence on
distance to wind farm
to detect navigational
error
Explicitly calculating
for waypoints
Includes effects of
AIS
Includes effects of
VTM
Models TSS
Calculations for
GL KF 1)
yes
GL new 2)
yes
MARIN KF 3)
unknown
MARIN
new 4)
unknown
yes
DNV 5)
yes
SSPA
yes
yes
yes
yes
ship type and
size
dependent
sea area and
vessel
size/type
dependent
yes
yes
yes
yes
no
yes
yes
yes
unknown
ship type
dependent
unknown
ship type
dependent
yes
yes
yes
yes
no
yes
no, but fishing
vessels
included
separately
no
time
dependent,
about 1E-10
no
yes
no, but
fishing
vessels
included
separately
yes
0.0003
unknown
unknown
0.0003
0.0003
within 20
minutes
sailing time
no detection
exp-function
no
no
exp-function
expfunction
no
no
no
no
yes
no
yes
yes
no
yes
unknown
no
no
yes
no
yes
unknown
no
yes
yes
yes
unknown
yes
every power
plant
every power
plant
every power
plant
yes
every
power
plant
unknown
park area
dependent on
ship type,
ship size,
load
condition
and speed
over ground
dependent on
ship type, ship
size, load
condition and
speed over
ground
dependent on
ship type, ship
size and load
condition
dependent
on ship
type, ship
size and
load
condition
unknown
dependent
on ship type
and load
condition
dependent on
ship type,
ship size and
load
condition
dependent on
ship type, ship
size and load
condition
dependent on
ship type, ship
size and load
condition
dependent
on ship
type, ship
size and
load
condition
unknown
dependent
on ship type
and load
condition
yes
yes
no
unknown
unknown
possible
Drifting model
Wind induced forces
Wave induced forces
Current induced
forces
83
SSPA REPORT NO:
2005 4028
AUTHORS:
Tidal induced forces
Engine failure rate
(1/hour)
no
yes
yes
yes
0.0002
unknown
unknown
0.00025
function
unknown
constants,
dependent on
time
function
drift velocity
wind velocity
drift velocity
and ship size
wind
velocity
time
dependent
time and
weather
dependent
no
every power
plant
every power
plant
every power
plant
Self repair
Emergency anchoring
dependent on
Salvage tug
Calculations for
time and
weather
dependent
every
power
plant
unknown
ship type and
size
dependent
possible
function
function
0.00025
wind velocity,
distance to
wind farm
time and
weather
dependent
time and
weather
dependent
unknown
park area
wind
velocity
Consequences
empirical,
based on
Estimate of amount
empirical
empirical
empirical
empirical
based on
statistics
of oil spilled
statistics
no
no
possible
yes
no
possible
Oil drift model
Estimate of number
based on
no
no
empirical
empirical
no
of fatalities
statistics
Estimate of size of
based on
no
no
no
no
no
economic damage
statistics
Estimate of influence
by
no
no
no
no
no
on radar
simulation
no
no
no
no
yes
no
Cost-Benefit Analysis
1)
GL KF: The models used in GL’s original report for Kriegers Flak (Otto and Petersen 2003).
2)
GL new: Most recent information on GL’s models based on GL’s later reports for Kriegers Flak (Povel et al. 2004 and Otto
2004), the SAFESHIP project (SAFESHIP 2005 and 2006), the harmonisation process (Bundesministerium für Verkehr-, Bau
und Wohnungswesen 2005), other wind farm risk analyses (Neuhaus and Thrun 2003) and personal communication (Povel
2007).
3)
MARIN KF: The models used in MARIN’s original report for Kriegers Flak (van der Tak and Rudolph 2003).
4)
MARIN new: Most recent information on MARIN’s models based on MARIN’s latest report for Kriegers Flak (van der Tak
2005b), the SAFESHIP project (SAFESHIP 2005 and 2006), the harmonisation process (Bundesministerium für Verkehr-, Bau
und Wohnungswesen 2005), other wind farm risk analyses (Kleissen 2006) and personal communication (Koldenhof 2007).
5)
DNV: Information on DNV’s models based on the SAFESHIP project (SAFESHIP 2005), the harmonisation process
(Bundesministerium für Verkehr-, Bau und Wohnungswesen 2005) and wind farm risk analyses (Christensen 2007).
6)
It is assumed from the harmonisation process (Bundesministerium für Verkehr-, Bau und Wohnungswesen 2005) that the
cruising speed is ship type dependent in GL’s and MARIN’s new models. In SSPA’s model it is possible to implement sea area
and vessel size/type dependent cruising speed in future model versions.
GL’s and MARIN’s models calculate the probability of collisions for every wind
power plant separately. In the SSPA calculation approach the entire wind farm
area is considered as a navigational hindrance and all uncontrolled or unplanned
ship entrances into the farm area generates obvious collision/contact hazards. It is
difficult to predict how a crew will react when the ship drifts or sail into the park
area. Will they take the necessary steps to avoid a collision or not? What will
these steps look like? Is there a chance that there will not be a collision and how
probable is it that a collision can be avoided? Therefore it seems to be more
reasonable to calculate the frequency of drifting and powered ships reaching the
park area. However, to be able to compare the results the model of SSPA was
adjusted so that the calculations lead to a collision frequency with the single
power plants and not the whole park area (see model description in Appendix).
84
SSPA REPORT NO:
2005 4028
AUTHORS:
Initially there was an attempt to keep SSPA’s model for drifting collisions as
simple as possible. In SSPA’s initial risk modelling, various variables were
averaged and unified. However, the averaged drift speed that was used in this
simplified model leads to an infinite return period for Kriegers Flak. The fast
drifters (vessels with high superstructure in relation to their underwater lateral
area), which are the only vessels that reach the wind farm before the salvage tug
arrives, will, however, not be included if averaging of the drift speed is applied.
Other parameters in this simplified model were also found to restrict the
applicability of the model and during the progress of the project different
parameters have been modified and improved in order to make the calculation
model more accurate and generally applicable. Introduced modifications and
elaboration of more detailed calculation routines include, for example, the
influence of the wind on emergency anchoring, salvage tugs, etc.
5.2
5.2.1
Input data
Description of the wind farm and the ship traffic
Offshore Ostsee Wind AG is the company developing the Kriegers Flak I project.
The offshore wind farm will be located in the German exclusive economy zone
(EEZ) in the Baltic sea, about 32 km northwest of the island Rügen, about 35 km
east from the Danish island Moen and about 35 km south of the Swedish coast
around Trelleborg. The overall dimension of the planned area is approximately 27
km2. The wind park will have an extension of about 7.5 km x 6 km and include a
maximum of 80 power plants. The power plants will have a capacity between
3 MW and 5 MW. The water depths in the area are around 20 to 45 meters. A
permit was granted in April 2005 for the construction of the wind farm (Windpark
Kriegers Flak 2008).
In MARIN’s investigation, only a preliminary study of the navigational risk is
conducted and therefore the wind park coordinates are roughly assumed. The
coordinates of the 52 power plants used in MARIN’s calculations are presented by
van der Tak and Rudolph (2003) and are also shown in Appendix. For the
simulations of MARIN’s results with SSPA’s model, these 52 coordinates are
used. The original calculations of GL (Otto and Petersen 2003) included nine
different wind park configurations. One of them was modified in their later
reports, ending up with a farm consisting of 80 power plants (see Otto 2004). The
coordinates of the power plants are not explicitly stated in GL’s reports. However,
WindPRO calculations for Kriegers Flak from October 22nd, 2004, shows the
coordinates of 80 power plants. Since the shape of this farm looks the same as the
one presented in GL’s latest report, the coordinates from the WindPRO
85
SSPA REPORT NO:
2005 4028
AUTHORS:
calculation (see Appendix) are used for the simulations of GL’s results with
SSPA’s model.
The data for the calculations which concern the shipping lanes used by MARIN
differs from that used by GL. MARIN (in van der Tak and Rudolph 2003)
presents 20 lanes that stand for the main contribution to their calculated results for
drifting and powered collision for the year 2000 and 2010, respectively. In total
they give a contribution of more than 99% of the total collision frequency. For the
simulations of MARIN’s results with SSPA’s model, these four sets of lanes are
used (see Appendix). GL (in Otto 2003) presents the 10 lanes that stand for the
main contribution to their calculated results for drifting collision and the 6 lanes
that stand for the main contribution to their calculated results for powered
collision. For the simulations of GL’s results with SSPA’s model, these two sets
of lanes are used (see Appendix). The traffic flow of the lanes is not explicitly
stated by GL. The figures presented in Appendix are estimated from Otto (2004)
and Povel and Petersen (2004). The length of the lanes used in SSPA’s
simulations are the same as presented by MARIN and GL. It has not been checked
whether these lengths are following the criterias presented in SSPA’s model
description (see Appendix).
Further data concerning the shipping traffic are not given in the reports by GL and
MARIN and are therefore assumed in SSPA’s calculations based on statistics. For
example, the ship type distribution used in SSPA’s calculations for drifting
collision are based on general statistics for the Baltic Sea with the assumption that
50% of the tankers are in loaded condition and that 50% are in ballast condition
(see table below).
Tabel 5.3. Ship type distribution (%) used in SSPA’s calculations for drifting collision.
%
Bulk/
comb
5.52
Tankers
loaded
7.92
Tankers
ballast
7.92
Gas
1.27
Gen.
cargo
56.26
Container
Reefers
3.73
2.01
RoRo
11.61
Passenger
Others
3.52
0.24
From statistical data for the area, average length and breadth of the ships are
assumed by SSPA to be 150 m and 25 m, respectively. The average vessel speed
is assumed to be 15 knots based on values presented in Chapter 3.
In SSPA’s simulations of MARIN’s and GL’s results, the power plant diameter is
assumed to be 5 m and the distance between the power plants is estimated to be
600 m (based on 80 power plants on an area of 27 km2).
The figures below shows the wind farm’s location and an example of lanes used
by MARIN and GL, respectively. For more illustrations, see MARIN’s and GL’s
reports.
86
SSPA REPORT NO:
2005 4028
AUTHORS:
Figure 5.1. MARIN’s wind farm and shipping lanes for powered collision
(illustrated by SSPA’s calculation program).
Figure 5.2. GL’s wind farm and shipping lanes for powered collision (illustrated
by SSPA’s calculation program).
87
SSPA REPORT NO:
2005 4028
AUTHORS:
To further illustrate the ship traffic situation, the figure below shows an AIS-plot
for the area, provided by the Swedish Maritime Administration for this project.
Figure 5.3. Example of density plot of recorded AIS track plots around the
Kriegers Flak 1 Site – 15 July 2006. (processed by Swedish Maritime
Administration (2006) using Gatehouse RAIS software).
5.2.2
Climate
Weather data, ice data, etc. are mainly provided in a text description in GL’s and
MARIN’s reports. Of special interest for the study in question is the wind
statistics. The figures below show the wind statistics presented in MARIN’s and
GL’s original reports. These data have been used in SSPA’s simulations.
88
SSPA REPORT NO:
2005 4028
AUTHORS:
Wind distribution MARIN
0,3
0,3
0,25
0,25
0,2
0,2
Distribution
Distribution
Wind distribution MARIN
0,15
0,15
0,1
0,1
0,05
0,05
0
0
1
2
3
4
5
6
7
8
9
10
11
12
N
NE
E
SE
BF
S
SW
W
NW
Wind direction
Figure 5.4. Statistical wind distribution at Kriegers Flak according to MARIN (van der
Tak and Rudolph 2003).
Wind distribution GL
0,25
0,25
0,2
0,2
Distribution
Distribution
Wind distribution GL
0,15
0,1
0,15
0,1
0,05
0,05
0
0
N
1
2
3
4
5
6
7
8
9
10
11
12
NE
E
SE
S
SW
W
Wind direction
BF
Figure 5.5. Statistical wind distribution at Kriegers Flak according to GL (Otto and
Petersen 2003).
89
SSPA REPORT NO:
2005 4028
AUTHORS:
NW
5.2.3
Self repair and Emergency anchoring and salvage
In MARIN’s latest report for Kriegers Flak it is stated that harmonised
assumptions have been used, which is why these assumptions are also used in
SSPA’s simulation of MARIN’s calculation. The assumptions are as follows:
•
•
•
Engine failure rate: 2.5E-4 per hour (see Chapter 3)
Self repair function: see Chapter 3 and Figure 5.8
Probability of anchor failure: see SSPA’s model description in Appendix
(based on Proposal 2 in harmonised diagram in Chapter 3 or below)
Figure 5.6. Probability of anchor failure (SAFESHIP 2005).
In GL’s latest report for Kriegers Flak the harmonised assumptions are not
mentioned, which is why the anchor failure curve presented in their original
Kriegers Flak report is used in SSPA’s simulations (see Germanischer Lloyd in
figure above). In SSPA’s simulations the drift velocity has been translated to wind
velocity (see figure below).
1,2
Probability of anchor failure
1
0,8
0,6
0,4
0,2
0
1
2
3
4
5
6
7
8
9
10
11
12
Wind speed (BF)
Figure 5.7. Probability of anchor failure used in SSPA’s simulations of GL’s
calculations for Kriegers Flak.
90
SSPA REPORT NO:
2005 4028
AUTHORS:
Instead of the harmonised engine failure rate and self repair curve, the data
presented in GL’s original report are used.
Engine failure rate: 2E-4 per hour (Otto and Petersen 2003)
Self repair function: see GL in figure below
•
•
100
90
Probability of unsuccessful repair [%]
80
70
60
GL
50
Spouge
MARIN
40
30
20
10
0
0
10
20
30
40
Time after failure [h]
Figure 5.8. Probability of unsuccessful repair. GL: Otto and Petersen (2003).
MARIN: SAFESHIP (2005) and Bundesministerium für Verkehr-, Bau und
Wohnungswesen (2005). Spouge: Spouge (1999).
Spouge (1999) has also presented frequencies of breakdowns for single engine
ships (see table below). He assumes engine failure rates which are independent of
the ship types and propulsion systems. He states that the rate can be divided into
three categories, which are shown in the table below. The first column represents
the average time which is needed until the failure is repaired and the crew has
taken control of the vessel again. The last category requires tug boat assistance
and repair in port. Spouge’s values are illustrated in the figure above.
91
SSPA REPORT NO:
2005 4028
AUTHORS:
Table 5.3. Frequencies of breakdowns for single engine ships (Spouge 1999).
Category
Frequency (per hour)
20 minutes
1.5 x 10-4
2 hours
4 x 10-5
2 days
1.5 x 10-5
In SSPA’s simulations for both MARIN and GL it is assumed that the water depth
and the sea bed condition do not restrict the possibilities for emergency anchoring.
The calculations presented from MARIN and GL do not include effects of
emergency salvage, which is why this possibility is also excluded from SSPA’s
simulations.
The drift velocity due to wind and waves is suggested to have a maximum value
of 4 knots, which has been agreed on by a group of experts (Bundesministerium
für Verkehr-, Bau und Wohnungswesen 2005).
5.2.4
Position on shipping lane
In SSPA’s simulations for both MARIN and GL, the probability of being on
position x on the shipping lane (Px) is assumed to be:
Px = 1/300
For more information, see SSPA’s model description in Appendix.
5.2.5
Course offset
The standard deviation for the course offset was not explicitly stated in the reports
of MARIN and GL, except for the lane south-east of the wind farm containing
tanker traffic. Otto (2004) states that the standard deviation for this lane is 1.23
nautical miles. In the simulations of SSPA, the standard deviation for similar lanes
has also been chosen to be 1.23 nautical miles. The standard deviation for the
remaining lanes has been chosen according to the harmonised assumptions
presented in SSPA’s model description (see Appendix) and the values chosen are
presented in Appendix.
92
SSPA REPORT NO:
2005 4028
AUTHORS:
5.2.6
Course deviation
In GL’s model, the calculations are very sensitive to the assumptions made for the
standard deviation describing the lateral distribution and the distance of the
shipping lane to the offshore wind farm (see Chapter 3). MARIN overcomes the
problem by introducing an improved model with a function representing the
course deviation. Based on some assumptions a distribution for the course
deviation is chosen, where the courses -30, -20, -10, 0, 10, 20, 30 have the
probability of 0.05, 0.1, 0.2, 0.3, 0.2, 0.1 and 0.05 respectively (van der Tak and
Rudolph 2003). SSPA’s model uses instead a Gaussian distribution which seems
more realistic. The standard deviation has to be chosen for the course deviation.
Fitting a Gaussian distribution on these values, with the mean value (µ) assumed
to be zero, gives a standard deviation of 15 degrees.
In the simulation of MARIN’s calculations, SSPA has described the course
deviation with a Gaussian distribution (standard deviation 15 degrees) as
described above. For the GL-simulations no course deviation is assumed since
GL’s model does not include that (standard deviation 0.0001 degrees is used).
5.2.7
Causation factor and onboard crew reaction
MARIN’s model includes the probability of detecting the navigational error and
to take measures to avoid a collision. In SSPA’s model, the probability that the
crew onboard is not able to react in time to correct the navigational error is called
Preact(x). It is dependent on the distance between the wind farm and the position of
the ship (D), and is therefore modelled as dependent on the x-position on the
shipping lane. The figure below shows weightings for offshore platforms
presented by MARIN (all curves except for exp(-D/(6L))) (van der Tak and
Glansdorp (Year unknown)). As already mentioned, MARIN uses the
Navigational Error Rate (NER) instead of the causation factor.
Preact(x) = exp(-D/(6L)), where L = ship length, is suggested in the literature to be
used together with the causation factor (Fujii and Mizuki 1998). The function is,
however, derived for navigation on lanes with a bend passing bridge piers. The
parameter D in the formula stands for distance from bend to bridge. In the SSPA
model, it is assumed that this function could be used for offshore wind farms with
D equal to the distance between the wind farm and the position of the ship. In the
figure below the ship length (L) is assumed to be 150 m.
In SSPA’s simulations of MARIN’s calculations Preact(x) = exp(-0.2D1.5) is used
together with the causation factor since this is a more conservative approach than
using Preact(x) = exp(-D/(6L)). The harmonised value of the causation factor is
used, i.e. 3E-04. The numerical value of NER is unknown.
In the simulations of GL’s calculations, it is assumed that Preact(x) = 1 since GL’s
model does not include that probability factor. In GL’s latest report for Kriegers
93
SSPA REPORT NO:
2005 4028
AUTHORS:
Flak (Otto 2004) the value of the causation factor is not explicitly stated.
However, Povel et al (2004) state the causation factor to vary between 1.11E-04
and 1.165E-04. The causation factor used in SSPA’s simulations of GL’s
calculations is 1.165E-04.
1
0,9
0,8
0,7
EXP(-D/(6L))
P_React
0,6
EXP(-0.2 D^1.5)
EXP(-0.072 D^2)
0,5
EXP(-0.575 D)
0,4
1/D
1/D^2
0,3
0,2
0,1
0
0
1
2
3
4
5
6
Distance [nm]
Figure 5.9. Weightings for offshore platforms presented by MARIN (all curves
except for exp(-D/(6L))) (van der Tak and Glansdorp (Year unknown)).
Preact(x) = exp(-D/(6L)), where L = ship length, is suggested in the literature to be
used together with the causation factor (Fujii and Mizuki 1998). The ship length
(L) is assumed to be 150 m.
94
SSPA REPORT NO:
2005 4028
AUTHORS:
5.3
Results
The table below shows SSPA’s simulations of GL’s and MARIN’s calculations
for Kriegers Flak compared to corresponding results presented by MARIN and
GL. Data and assumptions for the simulations are presented in previous chapters.
Note that the effect of emergency salvage is not included in the calculations and
simulations presented in the table.
Table 5.4. SSPA’s simulations of GL’s and MARIN’s calculations for Kriegers
Flak compared to corresponding results presented by MARIN and GL. The effect
of emergency salvage is not included.
Company
MARIN
GL
SSPA
Latest report
Latest report MARIN1) GL2)
Year 2010
Year 2000
2010
2000
Return period,
67
578
71
10
drifting [years]
Return period,
330
218
305
77
powered [years]
Return period,
56
158
58
9
total [years]
1)
SSPA’s simulations of MARIN’s calculations.
2)
SSPA’s simulations of GL’s calculations.
5.4
5.4.1
Conclusions and Discussion
Drifting collision
The return period from SSPA’s simulation is about the same as the one calculated
by MARIN. When it comes to the simulation of GL’s calculation, the difference is
large. If the harmonised functions for self repair and emergency anchoring are
used instead of GL’s original functions, the return period increases from 10 years
to 51 years. In such a simulation it is only the wind farm configuration, the
shipping lanes and the wind distributions that differ from the simulation of
MARIN’s calculation. It is difficult to find an explanation for the difference
between GL’s return period and the simulations made by SSPA of their
calculations. One possible reason could be the drift velocity. In the SAFESHIP
2005 project it was stated that the drift velocities calculated by GL are lower than
the drift velocities used by MARIN, which is the explanation for considerable
differences in the collision frequencies. However, if the drift velocity in the
95
SSPA REPORT NO:
2005 4028
AUTHORS:
simulation by SSPA of GL’s calculation is divided by a factor of 25, the return
period increases from 10 years to 516 years. This result indicates that the drift
velocity is most likely not the only reason.
5.4.2
Powered collision
The return period from SSPA’s simulation is about the same as the one calculated
by MARIN. The return period calculated by GL is about 3 times higher than the
one simulated by SSPA. As described earlier, the models for powered collision
are sensitive to changes of certain parameters.
A sensitivity analysis regarding different standard deviations for the course offset
(i.e. standard deviations describing the lateral distribution of the ships) has been
performed for the two versions of SSPA’s model. One version is used for
simulating MARIN’s calculations and includes course deviation and onboard
crew reaction. The other one is used for simulating GL’s calculations and does not
include course deviation and onboard crew reaction. To be able to compare the
results, the same data regarding wind farm coordinates and shipping lanes are
used (MARIN’s data has been chosen). The standard deviations chosen in the
base case has been multiplied with a factor that varies from 0.25 to 1.5. The
results are presented in the table below. In SAFESHIP (2005), MARIN’s and
GL’s models were compared in a similar way (see Chapter 3).
Table 5.5. Sensitivity analysis regarding different standard deviations describing the
lateral distribution of the ships for two versions of SSPA’s model.
Sensitivity expressed as collision frequency divided by
collision frequency for the base case
All standard deviations are
multiplied with
1.5
1.0 (base case)
0.75
0.5
0.25
SSPA’s model: version MARIN
3.5
1.0 (base case)
0.42
0.28
0.25
SSPA’s model: version GL
4.2
1.0 (base case)
0.18
0.01
0.01
As indicated above, the SSPA model version used for simulating MARIN’s
calculations is less sensitive to the assumptions made for the standard deviation
describing the lateral distribution of the ships. However, there are reasons to
believe that it is sensitive to assumptions regarding Preact(x) instead. The table
below shows the results of SSPA’s simulations of MARIN’s calculation for
Kriegers Flak for year 2010 regarding powered collision with different Preact(x).
96
SSPA REPORT NO:
2005 4028
AUTHORS:
Table 5.6. SSPA’s simulations of MARIN’s calculation for Kriegers Flak for year
2010 regarding powered collision with different Preact(x).
Preact(x)
Return period [years]
1/D
49
exp(-0.072D2)
248
exp(-0.2D1.5)
305
exp(-D/(6L))
720
As described in SAFESHIP (2005) (see Chapter 3), following the model of GL,
calculations of powered collision are very sensitive to the location of the shipping
lanes and the assumptions made for the standard deviation for the course offset.
The sensitivity analysis above also indicates this regarding the standard deviation.
From the results of the latest report of GL it is notable that the main risk
contribution is generated from shipping lanes at large passage distances. Shipping
lanes which are far away from the wind park are shown to give a bigger
contribution to the powered collisions than the shipping lanes which are less than
5 nm away from the planned offshore wind park.
5.4.3
Historical empirical accident statistics
Historical accident statistics can be used to compare and validate the predicted
results with real values. The zero-alternative calculations may also be based on
statistics on total accident frequencies in the region under consideration for wind
farm establishment. Accident statistics for grounding/stranding in the Swedish
EEZ are presented in the table below and show the recorded grounding statistics
between 1985 and 2006.
Table 5.7. Recorded grounding of ships in the area Sandhammaren-Falsterborev
on the coast of Skåne. Original material on reported accidents provided by the
Swedish Maritime Safety Inspectorate (Sjöfartsinspektionen 2006) has been
processed and compiled (see Appendix).
Time period: 1985-01-01—2006-07-13 (approximately 21,5 years).
Powered Grounding
Drifting Grounding
No.
Return period
No.
Return period
(year/grounding)
(year/grounding)
Passing ships
Not possible to
3
7
1
calculate 2)
Not possible to
Ships calling at
4
5
1
1)
port
calculate 2)
1)
Ships calling at port in Ystad and Trelleborg in Sweden
2)
Only 1 grounding.
In the discussion about acceptance criteria, the zero-alternative calculations have
been found to be relevant. The question here might be, if it is more important to
97
SSPA REPORT NO:
2005 4028
AUTHORS:
look at the return period for a wind park independently or if it is more relevant to
discuss the total increase of grounding frequency and collision frequency
compared to the present-day situation.
5.4.4
Consequences
MARIN classifies the consequences into damage for the wind power plants,
damage for the ship, human incidents, and environmental damage. The
probabilities for an oil spill and the spill volume are based on the kinetic energy,
on statistics and very simplified assumptions for how the damaged power plants
fall (direction away from the ship - on the ship, number of people hit by the
falling turbine, etc.).
GL calculates the probabilities for an oil spill and the spill volume based on
empirical formulas. Mainly the oil spill from tankers is looked at.
In SSPA’s case study the consequences are not included.
5.4.5
Influence of Kriegers Flak II on the calculations
The wind park Kriegers Flak II in the Swedish EEZ and the wind park Kriegers
Flak III in the Danish EEZ will influence the collision frequency of Kriegers Flak
I. So far the only studies which have been performed were made independently.
Following the analysis by SSPA interference of the different parks will be present
and cumulative risk aspects must be considered. The navigational risk should be
looked at for all parks independently as well as with a combination of all parks.
According to the information available at SSPA the parks I and II will be situated
so close to each other that it will not be possible to sail between the parks. If this
is the case the two parks can be considered as one big park.
98
SSPA REPORT NO:
2005 4028
AUTHORS:
Recommendations and Discussion
6
RECOMMENDATIONS AND DISCUSSION
farms. In the chapters below, conclusions and recommendations in general
regarding risk assessment methodology and in detail regarding calculation models
are presented.
6.1
Risk assessment methodology
The list below contains important areas identified during the progress of this
research project. Recommendations for each area are also discussed.
•
•
•
•
•
•
Transparent calculation models
Cumulative effects
Relative comparison
Cost-Benefit
Risk reduction measures
Accident preparedness
It is important that the calculation models are transparent. The intention with the
model developed by SSPA (see Appendix) is that all information about the model
should be explicitly stated. This includes the model structure as well as the input
data. The importance of transparent calculation models are exemplified in the
Kriegers Flak case study (see Chapter 5) where different versions of the SSPA
model and also different input data are used to illustrate how the calculated
collision frequency (or return period) is affected. Harmonisation processes such as
the German one also requires transparency in order to give recommendations
about for example input data. Harmonisation can be a natural step to take when
the models are presented in detail. However, as shown in chapter 2, the conditions
in the different EU-member states vary a lot and each country may identify and
prioritise various safety aspects differently, and total harmonisation may be
difficult. The pilot site for this project, Kriegers Flak, may serve as an illustration
of the need for harmonisation and bilateral/international assessment discussions.
If several wind farms are planned in the area, cumulative effects on the risk should
be studied. This may require cooperation between different countries. One
example is the proposed Swedish and German parks at Kriegers Flak that are
close neighbours, but are processed separately without consideration of
99
SSPA REPORT NO:
2005 4028
AUTHORS:
cumulative effects, while other more distant wind farms on the German side are
considered from an interaction point of view with Kriegers Flak.
As illustrated in chapter 5, collision frequency models are sensitive to changes of
certain assumptions. Calculated results in absolute terms should therefore be
carefully interpreted. One way of doing this is to make relative comparisons
instead of using absolute values of acceptance criteria. If acceptance criteria
should be used, it should be stated for which type of calculation model and with
which input data these criteria are valid. One important relative comparison is a
zero-alternative discussion where the navigational risk in a specific area is
compared quantitatively with and without the presence of the wind park.
Comparative studies of the calculated collision frequency of different traffic lanes
can also be applied in order to identify which ones that stands for the largest
contribution.
Another way of relating the results of a risk assessment/analysis is to put it in an
economic context. Cost-benefit analysis is not included in this research project but
could be an interesting task for future projects (see next chapter). One way could
be to study the estimated risk in relation to the electricity production of the wind
farm.
Example of risk reduction measures are presented in chapter 4. Measures that are
associated with low economic costs should always be considered even if the
estimated risk is low. If the estimated risk is high, also more expensive measures
must be considered.
Accident preparedness includes various safety measures but should also be linked
to a control program. One of the objectives with establishing and follow a control
program is that the risk and safety issues will be continuously checked and
updated during the whole life time of the wind farm.
6.2
Calculation Models
The structure of the SSPA calculation model (see Appendix) is similar to other
models used for wind farms and offshore platforms. However, there are models
using simulations (e.g. Monte Carlo simulations) but in the SSPA model no
simulations are used since these make the model less transparent. It is
questionable whether simulations give more accurate results of a risk analysis.
The SSPA model is designed to be simple and transparent, which gives a good
prerequisite for explaining the physics behind the model. Especially the model for
drifting collisions is straightforward, based on geometry. The one for powered
collisions is associated with questions concerning how to model navigational
100
SSPA REPORT NO:
2005 4028
AUTHORS:
behavior and human error. This is common questions for developers of this type
of models.
Collision frequency models are in general sensitive to changes of certain
assumptions. They also contain an amount of uncertainties. Calculated results in
absolute terms should therefore be carefully interpreted. This is also valid for the
SSPA model and the aim is to be as clear as possible concerning
sensitivity/uncertainty. This openness makes the SSPA model more useful and
shows the way to improvements of the model. It has for example become obvious
during the progress of this research project that the function describing the
probability that the crew onboard is not able to react in time to correct the
navigational error (onboard crew reaction) needs to be further investigated
together with the causation factor. One way of doing this would be possible if the
processing of recorded AIS-data could be further developed (see next chapter).
The German harmonisation process has laid a basis for a common harmonised set
of parameters which should be used in risk calculations. However, one should be
attentive to that the process has a set of models as a basis and there may be
recommendations that are valid only for these models and can therefore not be
used commonly.
6.3
Follow up activities
Follow up activities to be included in future research projects includes:
•
•
•
•
•
further develop processing of recorded AIS-data in order to improve
model input data.
develop calculation models in order to study possible increase of ship-ship
collisions due to more congested traffic lanes after a wind farm
establishment.
study accident statistics more thoroughly in order to improve model input
data as well as providing a basis for a zero-alternative discussion.
develop criteria for cost-benefit analysis for offshore wind farms.
further develop calculation models describing the consequences of a
collision.
101
SSPA REPORT NO:
2005 4028
AUTHORS:
References
7
REFERENCES
Ali, M.A., and D. Zheng. 2003. Ship Impact Analysis Cape Wind Associates
Wind Park. Attachment 1 to Navigational Risk Assessment Cape Wind
Project Nantucket Sound. Report prepared for US Army Corps of Engineers,
Regulatory Division. Wellesley, Massachusetts.
Anatec UK Limited. 2002. Burbo Bank Offshore Wind Farm – Navigation
Risk Assessment. Report prepared by Anatec UK Limited on behalf of
SeaScape Energy Ltd. Ref. A1070-SS-CR-1.
BalticMaster. 2007. Offshore Windfarm development and the issue of
maritime safety, Case Study „Kriegers Flak“ I, II and III. In process, Version
13.05.2007. Available from: www.balticmaster.org
Biehl, F. 2007. Personal communication with F. Biehl, TU Hamburg-Harburg,
11th May 2007.
Biehl, F. 2004. Rechnerische Bewertung von Fundamenten von Offshore
Windenergieanlagen bei Kollisionen mit Schiffen. Technische Universität
Hamburg-Harburg, 22 Oktober 2004.
Biehl, F., and E. Lehmann. 2006a. Collisions of Ships with Offshore Wind
Turbines: Calculation and Risk Evaluation. In Offshore Wind Energy
Research on Environmental Impacts, eds. J. Koller, J. Koppel, and W. Peters,
Springer-Verlag, Berlin and Heidelberg GmbH & Co.
Biehl, F. and E. Lehmann. 2006b. Collisions of Ships and Offshore Wind
Turbines: Calculation and Risk Evaluation. Proceedings of OMAE 2006. 25th
International Conference on Offshore Mechanics and Arctic Engineering.
Hamburg, Germany.
BMT Renewables. 2005. Safety at Sea Demonstration Project D Deliverables
D6-D8 Report No: D6_8.01 Revision 1, September 2005.
www.safetyastsea.se
Bonn Agreement. 2007. Bonn Agreement Counter-Pollution Manual.
Available: http://www.bonnagreement.org
Boverket. 2003. Planering och prövning av vindkraftsanläggningar. ISBN:
91-7147-737-3. (In Swedish).
102
SSPA REPORT NO:
2005 4028
AUTHORS:
References
British Wind Energy Association. 2007. Offshore Wind Farms.
http://www.bwea.com/ukwed/offshore.asp
Brown, C. 2005. Offshore Wind Farm Helicopter Search and Rescue Trials
Undertaken at the North Hoyle Wind Farm. Report written for the Maritime
and Coastguard Agency, MCA Contract MSA 10/6/239. From:
www.mcga.gov.uk/c4mca/mcga-safety_information/nav-com/offshorerenewable_energy_installations.htm
Bundesministerium für Verkehr-, Bau und Wohnungswesen. 2005.
Genehmigungsrelevante Richtwerte für Offshore-Windparks. Bericht einer
Arbeitsgruppe, Bundesministerium für Verkehr-, Bau und Wohnungswesen:
Bonn, 14.03.2005. (Main document and Appendix 2, both in German).
Burgermeister, J. 2007. Spain to Allow Offshore Wind Farms. Published by
RenewableEnergyAccess.com, 3 August 2007. Website:
www.renewableenergyaccess.com
Butterfield, S., Ram, B., and W. Musial. 2007. US Offshore Technology
Overview. 2007 Wind Powering America Summit Proceedings. From:
http://www.eere.energy.gov/windandhydro/windpoweringamerica/wkshp_200
7_state_summit.asp
Christensen, C.F. 2007. Navigational Risk Assessment Rødsand 2 Wind Farm.
Prepared for Dong Energy. Report No. 64402600- REP – 01, Revision No. 2,
May 2007. Det Norske Veritas (DNV).
Christensen, C.F.; Andersen, L.W. and Pedersen, P.H. Year unknown. Ship
Collision Risk for an Offshore Wind Farm. http://www.rambollwind.com/PDF/Referencer/Ship_collision_Risk_R%F8dsand.pdf
Danish Energy Authority. 2005. Offshore Wind Power – Danish Experiences
and Solutions. Danish Energy Authority, Copenhagen, Denmark. Website:
www.ens.dk
Davidsson, G., Haeffler, L., Ljundman, B. och Frantzich, H. 2003. Handbok
för riskanalys. Räddningsverket, Sverige.
den Boon, H., Friis-Hansen, P., Soone Ravn, E., Povel, D., Dalhoff, P., Stein,
J., van der Tak, C., Frouws, K., and H. Just. 2005. SAFESHIP Reduction of
Ship Collision Risks for Offshore Wind Farms. Final Report, Deliverable 154. http://www.e-connection.nl/?lang=nl&nav=euprojecten&pag=safeship
103
SSPA REPORT NO:
2005 4028
AUTHORS:
References
Department of Trade and Industry (DTI). 2005. Guidance on the Assessment
of the Impact of Offshore Wind Farms: Methodology for Assessing the Marine
Navigational Safety Risks of Offshore Wind Farms. UK Department of Trade
and Industry, in association with Maritime and Coastguard Agency (MCA)
and Department for Transport. Available from: www.dti.gov.uk
Deutscher Nautischer Verein. 2004. Arbeitspapier „OffshoreWindenergieparks“ Stellungnahme Windkraft, Stand 17.02.2004 5:35.
Di Marcantonio, M. 2007. Personal communication 2007-06-06 via Email
with Marisa Di Marcantonio, MUMM, Royal Belgian Institute of Natural
Sciences.
Dirdal, H. 2007. E-mail contact with Harald Dirdal, Havgul AS, 2007-04-19.
DNV (Det Norske Veritas). 1999. Acceptance Criteria for Formal Safety
Assessment Based Ship Rules. DRAFT Revision 1, internal DNV project
c383184, April 1999.
DNV (Det Norske Veritas), George Washington University and Rensselaer
Polytechnic Institute and Le Moyne College. 1996. Prince William Sound,
Alaska, Risk Assessment Study.
Earth Policy Institute. 2006. Global Wind Power Expands in 2006.
www.earth-policy.org/Indicators/Wind/2006.htm
Energimyndigheten Available on:
http://www.energimyndigheten.se/WEB/STEMEx01Swe.nsf/F_PreGen01?Re
adForm&MenuSelect=907FED45068A4282C1257211002D8F70 Accessed 27
August 2007.
Enertrag 2007. Offshore Wind Farm Côte d’Albatre. Available from:
www.enertrag.com/cmadmin_2_1130_0.html
Eriksen, T. och Haugen, S. 2006. Personlig kommunikation. Safetec Nordic
AS. Norge.
ESS Group. 2003. Navigational Risk Assessment Cape Wind Project
Nantucket Sound. Report prepared for US Army Corps of Engineers,
Regulatory Division. Wellesley, Massachusetts.
European Wind Energy Association. 2007. Barriers must be removed for full
exploitation of offshore wind. Published 5 December, 2007, Berlin. Website:
www.ewea.org
104
SSPA REPORT NO:
2005 4028
AUTHORS:
References
Foley, G. 2007. E-mail communication with Gerry Foley, Coastal Zone Policy
Department of Communications, Marine and Natural Resources of Ireland
(DCMNR), 2007-04-18 and 2007-04-19.
Forsman, B., Hüffmeier, J. och Johansson, J. 2007. Kompletterande
beräkningar och uppdaterad riskbedömning och konsekvensanalys för
planerad vindkraftpark ”Skottarevet”. SSPA-rapport 2005 3757-3 version 2.
SSPA Sweden AB, Göteborg. (In Swedish).
Fowler, T. 2007. A gust of activity, Houston Chronicle.
Germanischer Lloyd. 2002. Richtlinie zur Erstellung von technischen
Risikoanalysen für Offshore-Windparks. Selbstverlag des Germanischer
Lloyd.
Gästhamnsguiden. 2007. Gästhamnsstatistik 2005.
http://www.gasthamnsguiden.se/
Fujii, Y. and Mizuki, N. 1998. Design of VTS systems for water with bridges.
In Proceedings of The International Symposium on Advances in Ship
Collision Analysis, Gluver, H. and Olsen, D. (eds), pp. 177-190. 10-13 May
1998, Copenhagen, Denmark. A.A. Balkema, Rotterdam, Netherlands.
Hammar, L. och van Berlekom, W. 2004. Riskanalys för planerad svensk
vindkraftspark vid Kriegers Flak i södra Östersjön.
SSPA-rapport 2004 3473-1. SSPA Sweden AB, Göteborg.
Haugen, S. 1998. An overview over ship-platform collision risk modelling. In:
Risk and Reliability in Marine Technology. C. Guedes Soares (ed),
A.A.Balkema/Rotterdam/Brookfield.
Howard, M. and C. Brown. 2004. Results of the electromagnetic
investigations and assessments of marine radar, communications and
positioning systems undertaken at the North Hoyle wind farm by QinetiQ and
the Maritime and Coastguard Agency. Navigation Safety Branch, Maritime
and Coast Guard Agency, Southampton.
Institute of Shipping Analysis, BMT Transport Solutions GmbH, and Centre
for Maritime Studies. 2006. Baltic Maritime Outlook 2006 Goods flows and
maritime infrastructure in the Baltic Sea Region.
International Maritime Organization. 2002. Guidelines for Formal Safety
Assessment (FSA) for Use in the IMO Rule-Making Process. MSC/Circ. 1023,
MEPC/Circ. 392. 5 April 2002. International Maritime Organization. London,
UK.
105
SSPA REPORT NO:
2005 4028
AUTHORS:
References
Johansson, J. 2007. Kollisionsfrekvens för fartygstrafik med vindkraftparker
till havs– studie av beräkningsmodeller. SSPA-rapport 4006 4154 och
4006 4155. SSPA Sweden AB, Göteborg.
Johansson, J. och Forsman, B. 2007. Riskbedömning för planerad
vindkraftpark i Hanöbukten. SSPA-rapport 2006 4142. SSPA Sweden AB,
Göteborg. (In Swedish).
Johansson, J.; Forsman, B. och Hammar, L. 2005. Riskbedömning för
planerad vindkraftpark Skottarevet i Kattegatt utanför Falkenberg. SSPArapport 2005 3757-1. SSPA Sweden AB, Göteborg.
Johansson, J., Forsman, B. och Hüffmeier, J. 2007a. Riskbedömning för
planerad vindkraftpark vid Finngrunden. SSPA-rapport 2006 4302. SSPA
Sweden AB, Göteborg. (In Swedish. Not public).
Johansson, J., Forsman, B. och Hüffmeier, J. 2007b. Riskbedömning för
planerad vindkraftpark vid Storgrundet. SSPA-rapport 2007 4506. SSPA
Sweden AB, Göteborg. (In Swedish. Not public).
Johansson, J. et al. Ongoing project. Ship collision frequency models in
connection to offshore wind farms (Working title). SSPA-report 4007 4681.
SSPA Sweden AB, Göteborg.
Kjerstad, N. 2007. E-mail contact with Norvald Kjerstad, Professor for
Nautical Science, Høgskolen i Ålesund (Responsible for the navigational
safety assessment of Havsul offshore wind farm project), 2007-04-19.
Kjerstad, N., 2006. Vurdering og simulering av navigasjonsforhold og
skipstrafikk ved vindmølleparken Havsul-II. Høgskolen i Ålesund.
Kjerstad, N., 2005. Vurdering og simulering av navigasjonsforhold og
skipstrafikk ved vindmølleparken Havsul-V, Prosjekt nr.: 62008. Høgskolen i
Ålesund.
Kleissen, F. 2006. NSW – MEP : Maritime and marine risk assessment of
calamitous (oil) spills. Report OWEZ_R_280_20_07_2006. Report prepared
by MARIN for Nordzee Wind, July 2006.
Koldenhof, Y. 2007. Personal communication. MARIN.
Kristiansen, S. 2005. Maritime Transportation- Safety Management and Risk
Analysis. Elsevier Butterworth-Heinemann.
Lützen, M. and Friis Hansen, P. Year unknown. Risk Reducing Effect of AIS
Implementation on Collision Risk. Maritime Engineering, Department of
Mechanical Engineering, Technical University of Denmark, Lyngby,
Denmark.
106
SSPA REPORT NO:
2005 4028
AUTHORS:
References
Magnusson, K. 2002. Vindkraftpark Fladen. Kompletterande riskbedömning.
SSPA-rapport 2002 2733. SSPA Sweden AB, Göteborg.
(Bilaga 6 till Göteborg Energi, 2002. Miljökonsekvensbeskrivning. Planerad
vindkraftpark Fladen. 2002-07-02.)
http://www.goteborgenergi.se/Om_Goteborg_Energi_Var_verksamhet_Ny_en
ergi_Vindkraft_Fladen_Miljokonsekvenser_DXNI-5321_.aspx
Maritime and Coastguard Agency. 2004. Marine Guidance Note MGN 275:
Proposed UK Offshore Renewable Energy Installations (OREI) – Guidance on
Navigational Safety Issues. Hydrography, Meteorology, & Ports Branch,
Southampton, UK.
Minerals Management Service. 2006. Technology White Paper on Wind
Energy Potential on the U.S. Outer Continental Shelf. Published by Minerals
Management Service, U.S. Department of the Interior. Report available from:
http://ocsenergy.anl.gov
Neuhaus, S. and H. Thrun. 2003. Technical Risk Analysis Offshore Wind
Energy Park Thornton Bank. Germanischer Lloyd, Report No. GL O-03-291
Rev. 1, 2003-02-14.
Otto, S. 2004. Offshore-Windpark Kriegers Flak. 2. Ergänzung zur
Risikoanalyse. Bericht Nr. ERD2004.155, Version 1/2004-08-26.
Germanischer Lloyd.
Otto, S. und Petersen, U. 2003. Offshore-Windpark Kriegers Flak. Technische
Risikoanalyse. Bericht Nr. ERI 2003.54, Version 2/2003-10-23. Germanischer
Lloyd.
Pottinger Geherty Environmental Consultants Ltd. 2007. Draft Terms of
Reference Proposed NaiKun Wind Farm. Report prepared for NaiKun Wind
Development Inc. Vancouver, BC, Canada.
Povel, D. 2006. Collision risk analysis for offshore structures and offshore
wind farms. In Proceedings of OMAE 2006: 25th International Conference on
Offshore Mechanics and Arctic Engineering. June 4-9, 2006, Hamburg,
Germany.
Povel, D. 2007. Personal communication. Germanischer Lloyd.
Povel, D., Otto, S. and U. Petersen. 2004. Offshore-Windpark Kriegers Flak.
Ergänzende Risikoanalyse - Zusatzbetrachtungen, risikomindernde
Maßnahmen. Bericht Nr. ERD2004.032, Version 1/2004-04-22. Germanischer
Lloyd.
107
SSPA REPORT NO:
2005 4028
AUTHORS:
References
Povel, D. and Petersen, U. 2004. Offshore Wind Farm Kriegers Flak II
Sweden. Collision Risk Analysis Results Summary. Report No. ERD 2004.182,
Version 2.2/2004-12-07. Germanischer Lloyd.
Pryor, S., Shahinian, M., and M. Stout. 2005. Offshore Wind Energy
Development in the Great Lakes: A Preliminary Briefing Paper for the
Michigan Renewable Energy Program.
Ram, B. 2004. Offshore Wind Developments in the United States:
Institutional Issues, Environmental Regulations, and Jurisdictions. Presented
at World Renewable Energy Congress VIII, Denver, Colorado, 28 August – 3
September, 2004.
Rambøll. 2000a. ELSAM Ship Collision at Horns Rev. May 2000.
Baggrundsraport nr 23. Rambøll, Virum, Danmark.
Rambøll. 2000b. SEAS Skibskollisioner ved Rødsand. Juni 2000
(Havmøllepark ved Rødsand, VVM-redegørelse. Baggrundsraport nr 21, Juli
2000). Rambøll, Virum, Danmark.
Randrup-Thomsen, S., Andersen, L.W., and J.K. Gaarde. Year unknown. Risk
of Oil Pollution due to Ship Collision with Offshore Wind Farms.
SAFESHIP. 2005. Reduction of Ship Collision Risks for Offshore Wind
Farms. Collsision Frequencies. Deliverable No. 6; Version 2.0, 2005-01-27.
Germanischer Lloyd AG, Maritime Research Institute Netherlands MARIN,
Technical University of Denmark (section Maritime Engineering).
Farms. State of the Art of Risk Models. Deliverable No. 5; Version 1, 2006-1027. Germanischer Lloyd AG, Germanischer Lloyd Wind Energy GmbH,
Maritime Research Institute Netherlands MARIN.
Safetec. 2002. Gunfleet Sands Wind Farm. Collision Risk Assessment. Report
prepared by Safetec UK Ltd for Enron Wind Gunfleet Ltd. Main report,
Doc. No.: 20-268-2550-1471 Rev 01, February 2002.
Safety at Sea. 2005. Demonstration Project D. Deliverables D2-D5. Report
No: D2 5.01 Revision 01. March 2005.
Sandkvist, J. och Hammar, L. 2002. Utgrunden 2. Riskbedömning
navigatoriska risker. SSPA-rapport 2002 2817. SSPA Sweden AB, Göteborg.
Sjöfartsinspektionen. 2006. Rapport från SjöOlycksSystemet. 2006-07-13.
108
SSPA REPORT NO:
2005 4028
AUTHORS:
References
Spouge. 1997. ”Risk criteria for use in Ship Safety Assessment”, Marine Risk
Assessment conference organized by Institute of Marine Engineers, May
1997, ISBN 0-907206-85-9.
Spouge, J. 1999. A Guide To Quantitative Risk Assessment for Offshore
Installations. Publication: 99/100. The Centre for Marine and Petroleum
Technology (CMPT).
Starling, M. 2007. Safety at Sea Trans-national procedures and guidelines for
offshore wind farm risk management. Demonstration Project D, Deliverable
D9, Report No. D9, Revision No. 01a, 03/2007.
Sveriges Hamnar. 2007. Hamnstatistik. Trafik och Ekonomi.
http://www.transportgruppen.se/templates/Page.aspx?id=1743
Swedish Maritime Administration. 2006. Density plots of recorded AIS track
plots 1 – 15 July 2006. Processed by the use of the Gatehouse RAIS software.
Talisman Energy. 2004. Beatrice Windfarm Demonstrator Project Scoping
Report. Talisman Energy Limited, Aberdeen, U.K. Report available from:
www.beatricewind.co.uk
van der Tak, C. und Rudolph, D. 2003. Gutachten zur verkehrlichen Eignung
von Seegebieten für die Errichtung von Windenergieparks in der Ostsee.
Bericht Nr. 18761.620/4; 21.11.2003. MARIN, Wageningen, Netherlands.
van der Tak, C. 2005a. Comparison of different collision frequency models.
Seminar: Maritime Safety of Offshore Wind Farms, Wageningen.
van der Tak, C. 2005b. Risk Assessment for Kriegers Flak based on
harmonized assumptions and AIS data. Report Nr. 18761.620/6; 8 April 2005.
MARIN, Wageningen, Netherlands. (Not public).
van der Tak, C. and Glansdorp, C.C. Year unknown. Ship Offshore platform
Collision Risk Assessment (SOCRA).
Windpark Kriegers Flak. 2008. Available on http://www.wind-projekt.de/ and
http://www.ofw-online.de/pages/projects/pro_kriegersflak.html Accessed 29
January 2008.
Winergy. 2007. “Winergy Projects”. Available on:
www.winergyllc.com/projects.htm. Accessed 17 August 2007.
Örestads Vindkraftpark AB. 2000. Lillgrund Vindkraftpark. Risikoanalyse af
besejlingsforholdene. Dokument nr. P-052671-01, Udgivelsedato 24 okt 2000.
COWI.
109
SSPA REPORT NO:
2005 4028
AUTHORS:
Appendix A – Summary information on existing offshore
wind farms
APPENDIX A – SUMMARY INFORMATION ON EXISTING
OFFSHORE WIND FARMS
Netherlands
Denmark
Netherlands
Sweden
Sweden
U.K.
Denmark
Sweden
Denmark
Denmark
Denmark
Denmark
Denmark
Ireland
Lely
Tunø Knob
Dronten
Bockstigen (Gotland)
Utgrunden
(Kalmar Sound)
Blyth
Middelgrunden
Yttre Stengrund
(Kalmar Sound)
Horns Rev
Rønland
Samsø
Frederikshavn
Nysted
Arklow Bank
U.K.
Rostock/
Germany
Kentish Flats
Barrow
4
Breitling
U.K.
U.K.
Dollard/
Germany
Scroby Sands
Emden
Japan
Hokkaido
North Hoyle
U.K.
Denmark
Vindeby
2
Country
Site
2006
2006
2005
2004
2004
2004
2003
2003
2003
2003
2003
2003
2002
2001
2.5 MW, 1 turbine
90 MW, 30 turbines in 4
rows
90 MW, 30 turbines
4.5 MW, 1 turbine
60 MW, 30 turbines
1.2 MW, 2 turbines
60 MW, 30 turbines
25.2 MW, 7 turbines
165.6 MW, 72 turbines
7.6 MW, 3 turbines
23 MW, line, 10 turbines
17.2 MW, 8 turbines total
160 MW, 80 turbines
10 MW / line, 5 turbines
40 MW, curved line, 20
turbines
2001
2000
2000
1998
1996
1995
1994
5 MW / two rows, 11
turbines
2 MW / single line, 4
turbines
5 MW / two rows, 10
turbines
16.8 MW / single line, 19
turbines
2.5 MW / cluster , 5
turbines
10.5 MW / cluster, 7
turbines of 1.5 MW
3.8 MW, 2 turbines
Total Capacity, lay-out
1991
Operation
Start
7 km
500 m
8 - 10 km
30 m
2.3 km
700 m
7 km (closest)
10 km
9 km
500 m
3.5 km
14 - 20 km
5 km
2 – 3 km
1 km
12 km
3 - 4 km
30 m
6 km
800 m
1.5 – 3 km
Distance to
shore
15 - 20
2
5 (average)
2
4-8
7 - 11
5
5 - 10
1
11 – 18
6 – 14
8
3-6
6
7 - 10
6
3-5
4–5
2.5 – 5
Water
Depth (m)
Offshore wind farms in operation or under construction as of November 2007
75
80
70
100
68
47
67
73.5
69
61
78 - 78.8
70
60
60
58
41.5
50
40.5
39
37.5
Hub height
(m)
Monopile
Like land-based
Monopile
Like land-based
Monopile
Monopile
Monopile
Monopile
Concrete caisson
"Gravity",
Monopile
Drilled monopile
Drilled monopile
Drilled monopile
Concrete caisson
"Gravity"
Driven monopile
Drilled monopile
Driven monopile
Concrete caisson
Driven monopile
Concrete caisson
Foundation Type
305,000,000
(expected)
231,144,000
15,000,000
284,000,000
3
3
200,000,000
95,000,000
547,145,000
21,131,000
78,906,000
66,244,000
629,966,000
30,000,000
88,760,000
36,900,000
8,300,000
36,700,000
13,459,000
4,000,000
10,309,000
Annual Output
(kWh / y)
1
8
2009
2008
End of
2008
2008
180 MW, 60 turbines
300 MW, 60 turbines
120 MW, 60 turbines
180 MW, 54 turbines (2
adjacent windfarms with 27
turbines each)
9 km
30 km
5 km
23 km
7 km
25 km
7 km
10 - 18 km
Offshore Wind Farms under construction as of November 2007
Offshore Wind Farms that have begun Operations as of November 2007
U.K.
Belgium
U.K.
Netherlands
19 - 24 m
45
2-8
18
80
80
59
115 to tip
88
83.5
70
Monopile
Monopile
Lattice tower
Monopile
Monopile
Under construction
Phase 1 start-up
expected in 2008
Construction begun,
expected to begin
operation 2009
Under construction
330,000,000
(expected to be
operational in March
2008)
Construction nearing
completion. Expected
to be operational in
2008
400,000,000
(expected – operation
began in 2007)
315,000,000
(expected – operation
began in 2007)
Notes:
1. Annual production figures for Danish offshore wind farms are from 2005 and were obtained from: Faktablad P4 Vindmøller på havet. May 2006,
Dansk Vindmølleforening
2. Information on North Hoyle from www.natwindpower.co.uk/northhoyle/statistics.asp
3. Annual operating data for Scroby Sands and Kentish Flats are for the year 2006 and were obtained from www.berr.gov.uk/files/file41543.pdf,
“Capital Grant Scheme for Offshore Wind Annual Report – January 2006 – December 2006 Executive Summary”
Data compiled from the following sources:
http://www.offshorecenter.dk/offshorewindfarms/
http://www.bwea.com/offshore/worldwide.html
http://www.offshorewindenergy.org/
http://home.wxs.nl/~windsh/offshoreplans.html
www.offshore-wind.de/page/index.php?id=4765&L=1
Key:
Robin Rigg
Thornton Bank
Lynn & Inner
Dowsing
7
110 MW, 48 turbines
2007 2008
Sweden
Lillgrund
Windpark Q7
10 MW, 2 turbines
2007
U.K.
90 MW, 25 turbines
108 MW, 36 turbines
2007
2007
U.K.
Netherlands
Burbo Bank
Beatrice Wind Farm
Demonstrator,
Moray Firth
6
Egmond Aan Zee
5
4.
5.
6.
7.
8.
Information on Barrow Offshore Wind from: www.bowind.co.uk/keyfacts.htm
Information on Egmond Aan Zee from: www.noordzeewind.nl/
Information on Burbo Bank from: www.dongenergy.com/burbo/project/technology.htm
Information on Windpark Q7 from: www.q7wind.nl
Information on Thornton Bank from: www.c-power.be/
Wind farms
Appendix A
in the world
Figur 2.1
Existing wind farms
Report 2005 4028
Name of park
Europe
60oN
oMoray Firth
oFrederikshavn
oBockstigen
oUtgrunden
oYttre Stengrund
oTunø Knob
oSamsø
oMiddelgrunden
oLillgrund
oHorns Rev
oRønland
56oN
oBlyth
oRobin Rigg
52oN
oBarrow
oEmden
oBurbo BankoLynn & Inner Dowsing
oNorth
Hoyle
oLely
oArklow Bank
oScrobyoEgmond
Sands
oDronten
oWindpark Q7
oThornton Bank
oKentish Flats
oVindeby
oNysted
oBreitling
48oN
44oN
6oW
0o
6oE
12oE
18oE
Wind farms
Appendix A
in the world
Figur 2.2
Existing wind farms
Report 2005 4028
Name of park
U.K.
60oN
oMoray Firth
58oN
56oN
oBlyth
oRobin Rigg
54oN
oBarrow
oBurbo Bank
oNorth Hoyle
oLynn & Inner Dowsing
oLely
oWindpark Q7
oScroby Sands
oDronten
oEgmond
oArklow Bank
52oN
50oN
oThornton Bank
oKentish Flats
9oW
6oW
3oW
0o
3oE
6oE
Wind farms
Appendix A
in the world
Figur 2.3
Existing wind farms
Report 2005 4028
Name of park
North Sea
o
58 N
57oN
oRønland
56oN
oHorns Rev
55oN
54oN
53oN
oLynn & Inner Dowsing
oEmden
oNorth Hoyle
oLely
oScroby Sands
oWindpark Q7
oEgmond oDronten
52oN
oThornton Bank
oKentish Flats
51oN o
0
2oE
4oE
6oE
8oE
10oE
Wind farms
Appendix A
in the world
Figur 2.4
Existing wind farms
Report 2005 4028
Name of park
Baltic
o
60 N
59oN
58oN
oFrederikshavn
57oN
56oN
oBockstigen
oUtgrunden
oYttre Stengrund
oTunø Knob
oSamsø
55oN
oMiddelgrunden
oLillgrund
oVindeby
oNysted
o
oBreitling
54 N
53oN o
10 E
12oE
14oE
16oE
18oE
20oE
Wind farms
Appendix A
in the world
Figur 2.5
Existing wind farms
Report 2005 4028
Name of park
Japan
48oN
44oN
oHokkaido
40oN
36oN
32oN
120oE
126oE
132oE
138oE
144oE
150oE
Appendix B – Accident Statistics from the Swedish
(in Swedish).
1
Typ
Fiskefartyg
Fiskefartyg
Annika
Sekstant
Evita
Fiskefartyg
Saga Star Passagerarfartyg
Passagerarfartyg
Götaland
Saga Star Passagerarfartyg
Namn
Dest
Händelse
Indien
3 Kåseberga
Skillinge
9095 Travemuende Trelleborg
3147 Tallinn
2
P g a spänningsbortfall till övervakningssystemet en kort stund slogs detta ut.
Alla huvudmaskiner trippade med "black out" som följd. P g a black out
förlorades styrningen och fartyget grundstötte. Inga skador kunde upptäckas.
Fiskebåten var på väg från Kåseberga till Skillinge när en sjö sköljde över
däcket, varvid ett nät följde med på utsidan och fångades av propellern.
Motorn stannade och fartyget drev upp på sandstranden där fartyget
övergavs.
Vid bogsering från Tallinn till Indien för upphuggning brast bogserkabeln och
SEKSTANT (S) drev på grund. S skall avlägsnas från platsen.
Fartyget grundstötte vid inloppet till Trelleborg. Orsaken till grundstötningen
8226 Trelleborg
Travemuende var att isdriften p g a tjockan ej rätt kunde uppskattas av befälhavaren.
Fartyget grundstötte då det back-manövrerade i rännan in mot Trelleborg. Vid
tillfället rådde hård oso-lig vind, tjockdrivande is fanns i rännan och sikten var
nedsatt till 50-75 m pga tjockan. Fartyget togs loss kl 11.47 med hjälp av egna
maskiner samt isbrytarassistans. Vid efterföljande dockning konstaterades
bottenskador och skador på propellerbladen. Det kan dock inte fastslås när
dessa skador inträffat. Orsaken torde vara siktförhållandena samt isdriften i
rännan.
5159 Sassnitz
Trelleborg
Sannolikt felnavigering. Fartyget är utrustat med radar, ap-navigator, ekolod
och autopilot. Navigeringen anges dock ha skett på en höft. Skepparen säger
sig vara medveten om att det fanns en uppskjutande sten. Båten fick ett hål i
fören och tog in vatten men grundsattes för att inte sjunka. Båten bärgades
sedermera och hålet i fören reparerades.
20 Kriegers flak Skåre
Brutto Avg
1
1
N55°21,0'
E13°09,0'
N55°21,8'
E13°04,4'
Hamn
1 (Trelleborg)
1 Fiskefartyg
N55°22,70'
E14°7,50'
Fiskefartyg
Bogserande
fartyg är
passerande
och tappar
objekt
1 (fiskefartyg).
Hamn
(Trelleborg)
Hamn
(Trelleborg)
N55°20,20'
E13°8,80'
N55°21,80'
E13°32,0'
1
Powered Drifting Kommentar
N55°21,5'
E13°9,2'
Plats
Valda uppgifter ur olycksrapporteringsmaterial från Sjöfartsinspektionen (2006) i området Sandhammaren-Falsterborev på Skånes sydkust,
kompletterat med kategoriseringen gjord i denna studie avseende powered respektive drifting grounding.
Tidsperioden är 1985-01-01—2006-07-13 (dvs ca 21,5 år). OBS: Tabellen fortsätter på nästa sida.
Fiskefartyg
Torrlastfartyg
Fiskefartyg
Passagerarfartyg
Torrlastfartyg
Kattegatt
Lobo
Albakor
Polonia
Winland
Peter Pan Passagerarfartyg
2240 Riga
29875 Swinoujscie
1898 Kaliningrad
1741 Rostock
11
31356
3
ALBAKOR (A) skulle gå genom Öresund och ut i Atlanten. Bryggan var
bemannad av en styrman och utkik. Söder om Trelleborg grundstötte fartyget
och tog in vatten. Besättningen evakuerades. Vakthavande styrman var vid
tillfället alkoholpåverkad och dömdes senare till två månaders fängelse. Sju
Nordatlanten veckor efter grundstötningen bärgades fartyget och bogserades till Ystad.
En besättningsman hörde "konstiga" ljud då fartyget passerade Ystad
angöring boj. Undersökning utfördes och intryckningar som bedömdes vara
gamla, hittades.
Ystad
P.g.a. felnavigering grundstötte fartyget. Sannolikt har trötthet medverkat till
händelsen då vakthavande hade arbetat 26 timmar under de två senaste
dygnen. Fartyget tog sig loss för egen maskin till hamn.
Goole
Lissabon
Det tyska passagerarfartyget fick bottenskador i Trelleborgs hamnområde.
Den danska fiskebåten grundstött och förlist syd om Skåre. Enligt uppgift från
Stockholm radio hade en livräddningskryssare observerat båten tjuvfiskande
utanför Skåre under dagen. Ärendet redovisat av polisen till
åklagarmyndigheten Malmö. Misstanke om brott mot sjölagen och
sjötrafikförordningen. Grundstötningsorsak okänd. Anm: Dansk Skibstilsyn
informerats om händelsen. Inga åtgärder kommer att vidtagas från deras
sida.
Befälhavaren somnade på vakt och fartyget grundstötte. Med hjälp av
bogserbåtar kunde fartyget dras av grundet.
1
1
1
N55°23,6'
E013°47,5'
N55°18,5'
E012°48,'
1
N55°21,80'
E13°4,0'
?
N55°22,60'
E13°2,0'
N55°19,4'
E013°15,8'
1
N54°21,0'
E13°10,0'
?
Passerande
Hamn
(Ystad)
Passerande
stort
fiskefartyg.
Passerande
Fiskefartyg
Hamn
(Trelleborg).
Powered
antages då
inga uppgifter
om
motorstopp
etc uppgetts.
4
Sjöfartsinspektionen. 2006. Rapport från SjöOlycksSystemet. 2006-07-13.
Referens
Appendix C – Case Study Wind Farm Kriegers Flak:
Some detailed information
1
Power plant coordinates, MARIN
deg
north
54
54
54
54
54
54
54
54
54
54
54
54
54
54
54
55
54
54
54
54
55
55
54
54
54
54
54
55
54
54
54
54
54
54
55
54
54
54
54
55
55
54
54
54
54
55
54
54
55
55
54
54
min
north
57.8448
58.0272
58.4574
58.2102
58.6404
59.0706
58.3926
58.8228
59.253
59.6832
58.1448
58.575
59.0052
59.4354
59.8662
0.2964
58.3272
58.7574
59.1882
59.6184
0.0486
0.4788
58.0794
58.5096
58.9404
59.3706
59.8008
0.231
57.8316
58.2618
58.6926
59.1228
59.553
59.9832
0.4134
58.4448
58.875
59.3052
59.7354
0.1656
0.5964
58.6272
59.0574
59.4876
59.9184
0.3486
59.2398
59.6706
0.1008
0.531
59.4228
59.853
deg
east
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
min
east
5.601
6.2826
6.348
6.9648
7.0302
7.0956
7.6464
7.7118
7.7778
7.8432
8.2626
8.3286
8.394
8.4594
8.5254
8.5908
8.9448
9.0102
9.0762
9.1416
9.207
9.273
9.561
9.627
9.6924
9.7578
9.8232
9.8892
10.1778
10.2432
10.3086
10.374
10.44
10.5054
10.5708
10.9248
10.9908
11.0562
11.1216
11.1876
11.253
11.607
11.6724
11.7384
11.8038
11.8692
12.3546
12.42
12.4854
12.5514
13.0362
13.1022
2
Power plant coordinates, GL
deg
north
54
54
54
54
54
54
54
54
54
54
54
54
54
54
54
54
54
54
54
55
54
54
54
54
54
54
54
54
54
55
55
54
54
54
54
54
54
54
54
54
54
54
54
54
54
54
54
54
54
54
54
54
min
north
58.8892
58.6715
58.4542
58.2370
58.0200
59.2712
59.0513
58.8313
58.6115
58.3920
58.1725
59.6980
59.4760
59.2540
59.0320
58.8098
58.5878
58.3662
58.1447
0.1340
59.8967
59.6728
59.4487
59.2247
59.0005
58.7767
58.5525
58.3283
58.1048
0.3888
0.1628
59.9368
59.7108
59.4848
59.2590
59.0330
58.8070
58.5810
58.3555
59.7127
59.4850
59.2572
59.0293
58.8015
58.5737
59.7248
59.4903
59.2660
59.0325
58.8107
59.7357
57.0250
deg
east
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
min
east
7.4020
7.5900
7.7778
7.9653
8.1528
7.9182
8.0982
8.2783
8.4583
8.6380
8.8177
8.4175
8.5895
8.7615
8.9335
9.1053
9.2733
9.4488
9.6203
8.9505
9.1050
9.2688
9.4327
9.5965
9.7602
9.9238
10.0875
10.2512
10.4145
9.5928
9.7485
9.9040
10.0597
10.2152
10.3707
10.5262
10.6815
10.8368
10.9920
10.8627
11.0098
11.1572
11.3042
11.4513
11.5983
11.6225
11.7845
11.9005
12.0420
12.1763
12.4228
5.1917
3
54
54
54
54
54
54
54
54
54
54
54
54
54
55
55
54
54
55
55
55
55
54
55
55
54
55
55
54
57.4420
57.0833
57.8852
57.6033
57.3213
58.3762
58.0902
57.8043
57.5182
57.6487
57.7425
59.4093
59.1030
0.4558
0.1458
59.8347
59.5273
0.4300
0.1333
0.3962
0.1683
59.9405
0.4133
0.1818
59.9692
0.4292
0.1978
59.9668
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
13
5.7040
6.0280
6.2922
6.5677
6.8433
6.8683
7.1310
7.3937
7.6563
8.4733
9.3633
12.6068
12.7795
12.8172
12.9795
13.1425
13.3033
13.6667
13.8083
10.4210
10.5683
10.7155
11.2052
11.3117
11.4607
12.0317
12.1622
12.2925
4
Shipping lanes used for drifting collision, MARIN 2000
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Coordinates
waypoint 1
5513
1417
5442.9
1249.6
5434.9
1218
5501.4
1313.6
5446.1
1334.3
5518.8
1307.7
5518.8
1307.7
5432.5
1343.8
5513
1417
5515
1251.9
5501.4
1313.6
5446.4
1243.7
5446.1
1334.3
5446
1300.3
5501.4
1313.6
5432.5
1343.8
5523.5
1346.9
5515
1251.9
5434.9
1218
5515
1251.9
Coordinates
waypoint 2
5447.5
1242.3
5512
1421
5518.8
1307.7
5445.1
1333.3
5501.4
1313.6
5436.8
1214.6
5432.5
1343.8
5518.8
1307.7
5446.7
1243.4
5501.4
1313.6
5516
1252.5
5512
1421
5447
1258.4
5445.1
1333.3
5432.5
1343.8
5501.4
1313.6
5447.5
1242.3
5456.9
1436.7
5523.5
1346.9
5512
1421
No. of
ships/year
14940
14858
4221
1227
1175
4184
1768
1768
99
1337
1254
95
4508
4584
110
79
28
1865
27
10643
Shipping lanes used for drifting collision, MARIN 2010
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Coordinates
waypoint 1
5513
1417
5442.9
1249.6
5434.9
1218
5518.8
1307.7
5513
1417
5518.8
1307.7
5432.5
1343.8
5501.4
1313.6
5446.1
1334.3
5446.4
1243.7
5515
1251.9
5501.4
1313.6
5446.1
1334.3
5446
1300.3
5501.4
1313.6
5432.5
1343.8
5515
1251.9
5434.9
1218
5523.5
1346.9
5515
1251.9
Coordinates
waypoint 2
5447.5
1242.3
5512
1421
5518.8
1307.7
5436.8
1214.6
5446.7
1243.4
5432.5
1343.8
5518.8
1307.7
5445.1
1333.3
5501.4
1313.6
5512
1421
5501.4
1313.6
5516
1252.5
5447
1258.4
5445.1
1333.3
5432.5
1343.8
5501.4
1313.6
5456.9
1436.7
5523.5
1346.9
5447.5
1242.3
5512
1421
5
No. of
ships/year
18402
18360
5250
5193
177
2141
2141
1084
1038
170
1178
1108
4267
4400
94
69
1739
21
18
10458
Shipping lanes used for powered collision, MARIN 2000
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Coordinates
waypoint 1
5513
1417
5515
1251.9
5446.1
1334.3
5501.4
1313.6
5513
1417
5432.5
1343.8
5446.4
1243.7
5501.4
1313.6
5434.9
1218
5523.5
1346.9
5501.4
1313.6
5518.8
1307.7
5432.5
1343.8
5512
1230.2
5446.1
1334.3
5456.9
1310.8
5432.5
1343.8
5442.9
1249.6
5457.4
1304.1
5442.9
1249.6
Coordinates
waypoint 2
5447.5
1242.3
5501.4
1313.6
5501.4
1313.6
5445.1
1333.3
5446.7
1243.4
5501.4
1313.6
5512
1421
5432.5
1343.8
5523.5
1346.9
5447.5
1242.3
5516
1252.5
5432.5
1343.8
5518.8
1307.7
5457.4
1304.1
5457.4
1304.1
5457.4
1304.1
5456.9
1310.8
5512
1421
5445.1
1333.3
5446
1300.3
No. of
ships/year
14940
1337
1175
1227
99
79
95
110
27
28
1254
1768
1768
15
6
9
9
14858
15
4557
Shipping lanes used for powered collision, MARIN 2010
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Coordinates
waypoint 1
5513
1417
5446.1
1334.3
5515
1251.9
5501.4
1313.6
5513
1417
5432.5
1343.8
5446.4
1243.7
5501.4
1313.6
5434.9
1218
5523.5
1346.9
5518.8
1307.7
5432.5
1343.8
5501.4
1313.6
5512
1230.2
5446.1
1334.3
5456.9
1310.8
5432.5
1343.8
5442.9
1249.6
5457.4
1304.1
5442.9
1249.6
Coordinates
waypoint 2
5447.5
1242.3
5501.4
1313.6
5501.4
1313.6
5445.1
1333.3
5446.7
1243.4
5501.4
1313.6
5512
1421
5432.5
1343.8
5523.5
1346.9
5447.5
1242.3
5432.5
1343.8
5518.8
1307.7
5516
1252.5
5457.4
1304.1
5457.4
1304.1
5457.4
1304.1
5456.9
1310.8
5512
1421
5445.1
1333.3
5446
1300.3
6
No. of
ships/year
18402
1038
1178
1084
177
69
170
94
21
18
2141
2141
1108
9
4
6
6
18360
9
4379
Shipping lanes used for drifting collision, GL
No.
1
2
3
4
5
6
7
8
9
10
Start of
Route
section
54°46.26N
55°15.36N
55°18.384N
55°18.384N
54°46.62N
54°35.82N
55°15.498N
54°35.82N
54°46.272N
55°25.578N
Start of
Route
section
12°43.68E
14°13.92E
12°38.694E
12°38.694E
12°43.56E
12°16.602E
12°51.864E
12°16.602E
12°43.656E
12°40.692E
End of Route
section
55°12.06N
54°46.62N
55°25.578N
55°15.498N
54°35.82N
54°46.26N
55°15.342N
55°20.502N
54°50.562N
55°31.866N
End of
Route
section
14°15.96E
12°43.56E
12°40.692E
12°51.864E
12°16.602E
12°43.68E
14°13.938E
13°8.502E
14°0.558E
12°42.54E
Traffic on
the route per
year
17607
14926
18000
19286
21232
21232
17766
9776
9391
18000
End of
Route
section
12°43.56E
13°15.24E
13°33.552E
14°15.96E
12°43.56E
12°30.66E
Traffic on
the route per
year
14926
2240
2240
16187
1420
16
Shipping lanes used for powered collision, GL
No.
1
2
3
4
5
6
Start of
Route
section
55°15.36N
55°15.498N
55°0.9N
54°46.26N
55°13.2N
54°45.312N
Start of
Route
section
14°13.92E
12°51.864E
13°15.24E
12°43.68E
14°16.02E
13°33.552E
End of Route
section
54°46.62N
55°0.9N
54°45.312N
55°12.06N
54°46.62N
55°7.392N
7
Details from SSPA’s simulation of MARIN’s calculation for powered
collision. Base case.
Shipping
lane:
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Shortest
distance
to the
park
[nm]:
2.5854
0.97619
1.0245
0.95931
3.2903
0.9371
4.1348
0.9371
1.8898
1.4818
4.8587
4.8587
1.0447
0.94483
0.818
0.54863
1.0275
7.5258
0.8409
12.0812
Length
of
shipping
lane
[nm]:
58.4185
20.1962
19.1475
20.6587
58.1576
35.0682
59.8115
35.0682
68.1982
49.8748
52.4619
52.4619
19.6906
25.1808
21.5335
3.9318
32.2597
58.2072
21.6649
6.7433
Assumed
standard
deviation
for course
offset [nm]: Traffic Separation [yes/ no]:
1.23
yes
0.35
no
0.35
no
0.35
no
1.23
no
0.35
no
1.23
no
0.35
no
0.75
no
0.75
no
1.5
no
1.5
no
0.35
no
0.35
no
0.35
no
0.35
no
0.35
no
1.23
yes
0.35
no
2
no
Details from SSPA’s simulation of GL’s calculation for powered collision.
Base case.
Shipping
lane:
1
2
3
4
5
6
Shortest
distance
to the
park
[nm]:
2.5748
0.95041
1.0016
3.9408
3.3667
1.4162
Length
of
shipping
lane
[nm]:
57.5041
20.5883
19.574
57.3038
57.6839
43.8635
Assumed
standard
deviation
for course
offset [nm]: Traffic Separation [yes/ no]:
1.23
yes
0.35
no
0.35
no
1.23
yes
1.23
no
0.75
no
8
Appendix D – SSPA Calculation Model for Collision
Ship – Offshore Wind Farm
APPENDIX D – SSPA CALCULATION MODEL FOR
COLLISION SHIP – OFFSHORE WIND FARM
1
Appendix D – SSPA Calculation Model for Collision
Ship – Offshore Wind Farm
SSPA Sweden AB conducts an analysis for two cases; drifting collision (with a
disabled ship) and powered collision (ramming ship). The schematic diagram
below illustrates the components of the risk analysis model used by SSPA. It is
worth mentioning that for some locations, distinction in different draughts should
be made, because bigger ships may ground before a collision occurs.
Future traffic forecasts
Fishery
statistics
FUTURE SHIP
TRAFFIC SITUATION
Marina/club
statistics
Processed AIS data
Port call statistics
Assume
relocation/
redirection
of lanes
Lane relocation
influences ship-ship
collision probability?
PRESENT SHIP
TRAFFIC SITUATION
FISHING
VESSELS
Identication
Ship
Identication
of shippingShipShiptype,
Identication
of
shipping
type,
Identification
lanes.Ship type, size,
of shipping
lanes.
of shipping
Traffictype,
flowsize,
N, size,Cat.n
lanes.
Traffic
flow
N,
Cat.3
lanes.
waypoints
Traffic
flow
N,size,Cat.2
waypoints
Traffic
flow
N
Cat.1
waypoints
per lane, waypoints
Qualitative consideration of
collision probability/consequences
External environmental conditions.
Wind speed/direction distribution
Current, Waves, Depth, Ice, Fog,
Seabed composition
Additional risk
contribution
Lanes
crossing wind
farm area?
Collision frequency / return
period “Ship-ship collision”
Powered
collision
Drifting Collision
CALCULATION OF
COLLISION PROBABILITY
Engine failure rate
Causation factor
Offset at a
certain position
PLEASURE
CRAFT
Drift start at a certain position with a
certain speed
Course
deviation
Self repair
Emergency
anchoring
Emergency
salvage
Successful?
Collision
candidate?
Collision frequency/ return period
“Powered collision”
Consequences
2
Sailing speed
of tug boat
Drifting towards wind farm?
Location of
tug boat
“Drifting collision”
Bollard pull
Risk Model for “Drifting Collisions”
1
RISK MODEL FOR “DRIFTING COLLISIONS”
For drifting collisions, SSPA Sweden AB uses a basic method to
estimate the probability that ships experiencing a breakdown (i.e. loss
of power, propulsion and/or steering) drift into the wind farm. The
model includes estimations of frequency of ship breakdown at specific
locations and also of effectiveness of mechanisms that help take control
of the vessel again. These mechanisms include emergency salvage, self
repair, and anchoring.
3
The frequency of vessels drifting from a shipping lane and colliding
with an object near the lane can be written as:
FCD = ∑ N i ⋅ Fdrift ⋅ Ti ⋅ PD1 PD 2 PD 3
i
where
FCD=
Ni =
Fdrift=
Ti =
PD1=
PD2=
PD3=
Collision frequency of drifting vessels (per year)
Number of vessels of ship type i in the area around the
object (vessels/year)
Frequency of breakdown (per hour)
Average time a vessel of ship type i spends in the area
to be considered for the calculation of the collision
frequency (hours)
Probability of the ship drifting towards the object
Probability of not receiving any effective external help
before a collision occurs
Probability of no collision avoidance by the ship
before a collision occurs (i.e. the crew is unable to stop
the drifting through self-repair, anchoring, etc.)
If there are several shipping lanes, the total collision frequency is the
sum of the collision frequencies of each lane.
The model is not applicable for estimating collisions during war
situations, or due to terrorist attacks or volitional/ targeted ramming.
1.1
Number of vessels and main shipping lanes
Maps with information from the Automatic Identification System (AIS)
are used to identify the main shipping lanes in the area around the
object. Histograms are used to estimate the number of vessels that sail
on each identified shipping lane. The data are also used to estimate the
breakdown of different ship types and ship sizes. For cases where data
are not available for the ship classes, statistics from harbours and
authorities are used.
The model is based on corresponding models for oil platforms. The
frequency of drifting collisions is calculated as a total for an area of
typically 20 nautical miles radius around the platform. The dimensions
of a wind farm are much bigger than those of an oil platform, which
implies that a larger radius could be required. However, for a single
wind power plant this is not true.
4
It has been agreed on by a group of experts (Bundesministerium für
Verkehr-, Bau und Wohnungswesen 2005) that the length of the
shipping lanes to be considered to contribute to the collision frequency
corresponds to a drifting time of 24 hours.
In SSPA’s model no such limits are introduced as described above.
Geographical limitations of the length of the shipping lanes are used
instead, e.g. land areas located between the wind farm and the lane.
1.2
Average time
The average time a specific vessel sails in the area to be considered for
the calculation of the collision frequency depends on the length of the
lane the ship is sailing on and on the speed of the vessel on this lane.
Average speeds are taken for the vessels on the different lanes.
1.3
Probability of drifting towards the object
The probability of drifting towards the object under consideration
varies with position on the shipping lane and the course of the shipping
lanes. For each point on the lane the wind farm is covered by an angle
which overlaps with the wind directions as shown in the example in the
figure below (division in 4 wind directions):
5
αS
αW
N
W
E
Sn
S
The probability is therefore dependent on the size of the angle, the
distance from the wind farm to the shipping lane and the frequency of
the wind blowing from the different directions. The factor PD1 is
calculated as follows:
N wd
 R x αw
PD1 = ∑  w
w=1  360 / N wd
where:
Nwd =
α w=
Rw =



number of divisions in different wind directions
angle which is covered by the wind farm in the wind
direction w
frequency for wind from direction w
(For the figure above, the angle is measured at the centre of Sn. In
SSPA’s calculation program, it is measured at the endpoints of Sn. This
approximation will most probably not have an influence on the
calculated results.)
1.4
External help (salvage tug)
External help for drifting vessels is assumed to be by emergency
salvage. This help depends on certain factors:
6
The salvage tug needs a certain time to respond, i.e. there needs
to be an available tug and it requires time to leave the harbour/
position.
The time for the salvage tug to reach the vessel depends on the
salvage tug position, vessel position, drift speed and salvage tug
speed. The sailing speed of the salvage tug depends on the sea
state and the wind velocity.
The time for the crews to connect a line. This time is strongly
dependent on the equipment of the two vessels involved as well
as the training of their crews.
The time to take control of the drifting vessel.
The performance of the tug, which is measured in tonnes of
bollard pull. The required power depends on the size and type
of the drifting vessel and the wind conditions.
•
•
•
•
•
When salvage tugs with different capabilities can reach the drifting
vessel in time, the one with the best performance is assumed to tow it.
It is assumed that a collision is avoided once the tug takes control of the
disabled vessel.
1.5
Self-repair, emergency anchoring, etc.
The frequency of breakdown (engine failure rate) for a single engine
ship is taken as Fdrift = 2.5x10-4 per hour (Bundesministerium für
Verkehr-, Bau und Wohnungswesen 2005). It is assumed that the
engine failure rate is independent of the ship type and propulsion
system.
One part of PD3 is the probability of no successful self-repair. This
probability depends on time it takes for the crew to repair the engine
failure without external help. A function has been derived based on
statistics from Dutch waters (additional details published in:
SAFESHIP 2005 and Bundesministerium für Verkehr-, Bau und
Wohnungswesen 2005), which depends on the time to repair:
f(t) = 1
f(t) =
1
(1.5 ⋅ (t − 0.25) + 1)
for t< 0.25 h
for t> 0.25 h.
where t = time after the engine failure occurred (hours)
PD3 therefore varies with the distance to the wind farm and the drift
speed of a certain vessel.
7
Another part of PD3 accounts for the cases when the crew is unable to
stop the ship with emergency anchoring. The probability for this is
connected to the water depth, the type of sea bottom, the wind velocity,
the drift speed, the distance from the wind farm and the ship size. The
figure below shows the probability of anchor failure for different wind
speeds for typical sea bottom characteristics of the Baltic Sea, which
has been agreed on of a group of experts (Bundesministerium für
Verkehr-, Bau und Wohnungswesen 2005). If the anchor holds, the
anchor prevents the drifting ship from reaching the wind farm.
1
Probability of anchor failure [%]
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0
2
4
6
8
10
12
Wind speed [BF]
1.6
Drift velocity
The drift velocity of the disabled vessel is assumed to be constant for
the calculations. It is modelled by the energy equation that shows the
sum of all forces acting on the ship to be equal to zero.
FResistance + FWind + FWave + FCurrent = 0
where
8
FResistance=
Resistance of the ship through the water
FResistance =
1
2
ρ water LiTil Cd _ i vdrift
_ il
2
FWind=
FWind
wind force acting on the ship
1
2
= ρ air AL _ il CdWind _ i vWind
2
FWave=
wave force acting on the ship (assumed to be FWave=½ FWind)
FCurrent=
force acting on the ship due to current
vdrift _ il =
drift velocity of ship of type i in loading condition l in wind and
waves at different wind velocities
wind velocity
vwind =
ρ air =
ρ water =
density of air (1.35 [kg/m3])
ALil =
density of water (1025 [kg/m3])
lateral wind surface of ship of type i in loading condition l
Til =
draught of the ship of type i in loading condition l
Li =
length of the ship of type i
CdWind _ i =
lateral wind resistance coefficient of the ship of type i
Cd _ i =
lateral resistance coefficient of the underwater body of the ship of
type i
The coefficients and parameters are based on data from SSPA Sweden
AB and on published approximations. The force acting on the ship due
to current is most of the time assumed to be negligible for the Baltic
Sea. The drift direction is assumed to be the same as the wind direction.
1.7
Sailing into the park versus hitting a wind power
plant
The model described above calculates the frequency of disabled ships
drifting into the wind farm area. It is also possible to calculate the
frequency of collisions of disabled ships with the single power plants.
For this case, the results calculated for reaching the park area should be
multiplied by the following expression:
(1-(1-(L+D)/b)r)
9
where
L = ship length
D = power plant diameter
b = distance between power plants
r = number of rows of power plants
10
Risk Model for “Powered Collisions”
2
RISK MODEL FOR “POWERED COLLISIONS”
The model used at SSPA Sweden AB estimates the probability of ships
being navigated incorrectly into the wind farm. The incorrect
navigation can be caused by human, technical and/or watch keeping
failure.
The collision frequency for powered collisions on a shipping lane is
calculated using the following equation:
FCP = ∑
x
∑ ∑
N ⋅ Px ⋅ Poffset ⋅ Pcourse ⋅ PC1 ⋅ PC 2 ⋅ PC 3 ⋅ Preact ( x)
offset course
where
FCP=
N=
x=
Px =
Poffset =
Pcourse =
PC1 =
PC2 =
PC3 =
Preact(x) =
Frequency of a passing ship colliding under power (per
year)
Total traffic on the shipping lane (vessels/ year)
Position on the shipping lane
Probability of being on position x on the shipping lane
Probability of having a certain offset on the current xposition (Gaussian distribution + uniform distribution)
Probability of following a certain course heading
towards the object (Course deviations are assumed to
follow a Gaussian distribution.)
Probability of human failure during planning and
execution of the passage of an object
Probability of technical failure of navigational
equipment or of watch keeping failure due to factors
such as lack of attention during lookout on the bridge
or bad visibility
Probability of failure of the wind farm safety
equipment/ crew or a potential stand-by boat to warn
the passing ship in time to avoid a collision
Probability of the crew onboard being unable to react in
time to correct the navigational error (dependent on x)
If there are several shipping lanes, the total collision frequency is the
sum of the collision frequencies of each lane.
The model is not applicable for estimating collisions during war
situations, or due to terrorist attacks or volitional/ targeted ramming.
11
Risk model Powered
collision (/grounding)
Shipping
lane
Powered collision/grounding is a
result of mistaken position that will
be manifested as an offset from
desired track on lane and or a
result of mistaken heading that will
be manifested as a deviation from
desired course direction.
Wind farm
Way point
Safety
zone
500 m
Way point margin
The offset and deviations may be
caused by human and/or technical
failures. Depending on the
distance to the obstacle, the crew
may be able to detect the error and to
correct the mistake.
Geometric collision candidates are
found by their position on the shipping
lane, their offset from the shipping lane
and their course over ground.
Offset and deviations vary
independently and collision
candidates are identified by
multiplying the two independent
Gaussian distributed variables.
Course offset
normal distribution +
2 % homogenous distribution
The chance of detection and
correction of the collision course is
a function of the time/distance
remaining to the collision position and
on how often the position is checked.
p
Course deviation
0.3
0.2
0.1
0.05
-30 -20 -10 0 10 20 30 deg
12
2.1
Number of vessels and main shipping lanes
It is assumed that parts of the shipping traffic follow certain routes.
This is modelled by shipping lanes on which the ships sail. To identify
the main shipping lanes in the wind farm surroundings and to estimate
the traffic on these lanes, AIS plots and histograms are used in the same
way as for drifting collisions (see previous chapter). If the current
traffic sails through the area of a future wind farm, the traffic is moved
according to certain assumptions. The length of the shipping lanes are
suggested to correspond to a traffic area that measures about 15
nautical miles (or 20 nautical miles) from the outer corners of the wind
farm, which has been agreed on by a group of experts
(Bundesministerium für Verkehr-, Bau und Wohnungswesen 2005). In
SSPA’s model no such limit is introduced. Geographical limitations of
the length of the shipping lanes are used instead, e.g. land areas located
between the wind farm and the lane.
2.2
Position on shipping lane
The shipping lanes are split into a number of parts and the calculations
for each part is summarised. This is an approximation to integrating
over the whole lane. The probability of being on position x on the
shipping lane (Px) is assumed to be equally distributed over the
shipping lane as follows:
Px = 1/ nsplit
where nsplit is the number of parts the lane has been split into in the
calculation. This assumption presupposes that the sailing speed is
constant over the shipping lane and that each part of the lane has the
same length. The original function is
Px = tx / ttot
where ttot is the total time it takes to travel from the start to the end of
the shipping lane and tx is the time it takes to travel from the start to the
end of the part of the shipping lane representing the position x.
13
2.3
Course offset
The lateral distribution on the lanes is assumed to usually follow a
Gaussian distribution. The standard deviation (σ) for the Gaussian
distribution is estimated from histograms. If the standard deviation
cannot be estimated from histograms because, for example, the lane has
been moved or certain parts of the shipping lane have special
distributions, reference values from the table below have been used.
The source of the table is: Bundesministerium für Verkehr-, Bau und
Wohnungswesen (2005).
Description
Standard deviation for Gaussian
distribution [nm]
0.2 to 0.3
0.3 to 0.4
Port approach
Conspicuous navigational points, e.g.
navigational marks, buoys
separation
0.5
0.5 to 1.0
2.0
The mean value (µ) of the Gaussian distribution is usually assumed to
be zero. If no further information is available, SSPA assumes that the
width of the Gaussian distribution used in the calculations is taken as
12 times the standard deviation (i.e. 12σ).
Part of the shipping traffic does not follow a Gaussian distribution. For
this part of the traffic a uniform distribution is assumed. If no further
information is available, the width of the uniform distribution is taken
as 6 times the standard deviation (i.e. 6σ). The uniform distribution is
then taken as 2% of the normal distribution (Bundesministerium für
Verkehr-, Bau und Wohnungswesen 2005).
Accordingly, the Gaussian distribution only partly describes the
probability of having a certain course offset along an axis perpendicular
to the shipping lane, here called PoffsetG. The other part, the uniform
distribution, of Poffset is PoffsetU, i.e.
Poffset = PoffsetG + PoffsetU
As an example, the two figures below illustrate PoffsetG and PoffsetU,
respectively, for a special interval instead of for a certain point along
the axis. The probability of having an offset in this interval is called Fd
in this model description.
14
The first figure shows the Gaussian distribution on the shipping lane
and the possible collision candidates who are at risk of a collision with
an object (FdG), if they continue their course straight ahead (i.e. if the
course deviation is zero degrees). As the figure shows, the distance
between the lane and the object also influences FdG.
FdG
The second figure shows the uniform distribution on the shipping lane
and the possible collision candidates who at risk of a collision with an
object (FdU), if they continue to sail on their course straight ahead (i.e.
if the course deviation is zero degrees). As the figure shows, the
distance between the lane and the object also influences FdU.
∆
δ
FdU
15
The total proportion of vessels that are in the part of the lane directed
towards the wind farm if the course deviation is zero degrees is
assumed to be:
Fd = FdG + FdU
where FdU = 0.02 · δ / 6σ
and δ ≤ ∆
2.4
Course deviation
In SSPA’s model it is assumed that the course deviation of a ship can
vary between -90 and 90 degrees, i.e. no course deviations leading to
ships sailing in the opposite direction are possible.
The course deviation is assumed to usually follow a Gaussian
distribution. The standard deviation (σ) for the Gaussian distribution is
difficult to estimate. Further studies of AIS data are needed. To date,
figures from MARIN have been used in SSPA’s model as a base for
estimation. MARIN does not use a Gaussian distribution for the course
deviation. They use a distribution where the courses -30, -20, -10, 0,
10, 20, 30 have the probability of 0.05, 0.1, 0.2, 0.3, 0.2, 0.1 and 0.05
respectively (van der Tak and Rudolph 2003). Fitting a Gaussian
distribution to these values, with the mean value (µ) assumed to be
zero, gives a standard deviation of about 15 degrees.
2.5
Causation factor
Different values of PC1, PC2 and PC3 are presented in the literature (see
e.g. Spouge 1999). A combined factor PC is often used. PC (Causation
probability, i.e. probability of failure to avoid an obstacle on the
navigation route) has been discussed extensively in the literature since
the seventies. Most estimates of PC have been based on data available
for groundings and ship-ship collisions. Two approaches published in
1974 constitute the basis for most of the other estimations: Fujii’s or
MacDuff’s (Larsen 1993). The International Association of Marine
Aids to Navigation and Lighthouse Authorities (IALA) also refers to
Fujii for the value of PC (IALA 2007). Ramböll (2000) suggests
PC = 2 x 10-4 be used based on Fujii’s estimations. GL and DNV have
agreed to use the causation factor of PC = 3 x 10-4 for a ship not taking
corrective action when on collision course (SAFESHIP 2005). In SSPA
16
Sweden AB’s risk model for powered collisions this factor is assumed
to be valid (PC = 3 x 10-4).
There are also other factors influencing PC, such as bad visibility.
Larsen (1993) and Spouge (1999) illustrate such factors.
Onboard crew reaction
In SSPA’s model, the probability that the crew onboard is not able to
react in time to correct the navigational error is called Preact(x). It is
dependent on the distance between the wind farm and the position of
the ship (D), and is therefore modelled as dependent on the x-position
on the shipping lane. The figure below shows weightings for offshore
platforms presented by MARIN (all curves except for exp(-D/(6L)))
(van der Tak and Glansdorp (Year unknown)). As already mentioned,
MARIN uses the Navigational Error Rate (NER) instead of the
causation factor. Preact(x) = exp(-D/(6L)), where L = ship length, is
suggested in the literature to be used together with the causation factor
(Fujii and Mizuki 1998). The function is, however, derived for
navigation on lanes with a bend passing bridge piers. The parameter D
in the formula stands for distance from bend to bridge. In the SSPA
model, it is assumed that this function could be used for offshore wind
farms with D equal to the distance between the wind farm and the
position of the ship. In the figure below the ship length (L) is assumed
to be 150 m.
A more conservative approach is to use for example
Preact(x) = exp(-0.2D1.5) together with the causation factor instead of
using Preact(x) = exp(-D/(6L)).
1
0,9
0,8
0,7
P_React
2.6
EXP(-D/(6L))
EXP(-0.2 D^1.5)
0,6
EXP(-0.072 D^2)
0,5
EXP(-0.575 D)
0,4
1/D
1/D^2
0,3
0,2
0,1
0
0
1
2
3
Distance [nm]
17
4
5
6
2.7
Sailing into the park versus hitting a wind power
plant
The model described above calculates the frequency of ships sailing
into the wind farm area as the result of an error. It is also possible to
calculate the frequency of ships colliding (under power) with the single
power plants. For this case, the results calculated for reaching the park
area should be multiplied by the following expression:
(1-(1-(B+D)/b)r)
where
B = ship breadth
D = power plant diameter
b = distance between power plants
r = number of rows of power plants
18
References
3
REFERENCES
Bundesministerium für Verkehr-, Bau und Wohnungswesen. 2005.
Genehmigungsrelevante Richtwerte für Offshore-Windparks. Bericht einer
Arbeitsgruppe, Bundesministerium für Verkehr-, Bau und Wohnungswesen:
Bonn, 14.03.2005. (Main document and Appendix 2, both in German).
Fujii, Y. and Mizuki, N. 1998. Design of VTS systems for water with
bridges. In Proceedings of The International Symposium on Advances in
Ship Collision Analysis, Gluver, H. and Olsen, D. (eds), pp. 177-190. 10-13
May 1998, Copenhagen, Denmark. A.A. Balkema, Rotterdam, Netherlands.
IALA. 2007. IALA Risk Management Tool. The IWRAP Model. Presenterat
av: Mr Ian Gillis, Risk Assessment Officer, Canadian Coast Guard.
http://www.iala-aism.org/
Larsen, O. D., 1993. Ship Collision with Bridges. The Interaction between
Vessel Traffic and Bridge Structures. International Association for Bridge
and Structural Engineering (IABSE).
Ramböll. 2000. SEAS Skibskollisioner ved Rödsand. Juni 2000
(Havmöllepark ved Rödsand, VVM-redegörelse. Baggrundsraport nr 21, Juli
2000).
Farms. Collsision Frequencies. Deliverable No. 6; Version 2.0, 2005-01-27.
Germanischer Lloyd AG, Maritime Research Institute Netherlands MARIN,
Technical University of Denmark (section Maritime Engineering).
Spouge, J. 1999. A Guide To Quantitative Risk Assessment for Offshore
Installations. Publication: 99/100. The Centre for Marine and Petroleum
Technology (CMPT).
van der Tak, C. and Rudolph, D. 2003. Gutachten zur verkehrlichen Eignung
von Seegebieten für die Errichtung von Windenergieparks in der Ostsee.
Bericht Nr. 18761.620/4; 21.11.2003. MARIN, Wageningen, Netherlands.
19

Methodology for Assessing Risks to Ship Traffic from

Transcription

Similar documents

1. After making contact with the village Bolbo, the `Peace` sailed on

Untitled

USS CONYNGHAM DDG-17 History Collection

ordering instructions mail order

WITH BROADPOINT

Conceptual design of offshore concrete structures

Christopher F. Wood USCGC CHASE ship roster.pub

Activity - Scholar

Celigo - Calcey Technologies

mystery ship of wwii, the surcouf